GLiDE v1.2 Decision Model Results

GLiDE v1.2

Explore GLiDE v1.2: exact decision configuration, published specifications, source-linked benchmarks and API tariffs when available.

Latest versionPrevious version: GLiDE (September 2026)Release source ↗

Published specifications

Provider: Fastino
Version: fastino/GLiDE v1.2 (one forward pass)
System type: Trained checkpoint
Access: Open weights
License: Fastino Community License
Inputs: text, image
Decision types: Choice, Yes / no, Score
Published size: 27B
Active parameters: Not published
Context / input limit: 40,000 tokens per question prompt by default; hosted rendered prompt limit 65,536
Announced: 2026-10-22

Hugging Face · Announcement · Website

Model notes

Post-trained from an open-weight 27B-class hybrid with 48 Gated DeltaNet and 16 full-attention layers. Hugging Face counts 26,085,330,160 tensors, the community board shows 27.8B, and Fastino's docs say 27B, shown here. The repository reports gated false while its card says download needs a token with read access to the fastino org. Card latency, bf16, warm, one request at a time: median 84 ms on an RTX PRO 6000, 48 ms on a B200. Fastino says its hosted API runs v1.2. System 1 v2's Fastino GLiDEv2.3 BF16, S1MB's fastino/glide and Benchmark Heaven's hosted API row are not mapped here, because no primary source ties those labels to this checkpoint. GLiDE v1.1 is not registered. Typed outputs are probabilities over supplied options; confidence is not a correctness guarantee.

Checkpoint revision: d199b6d7e04523d87986b8b68cd62272ed1cefc0

← All decision models · Compare this decision system →

Published API tariff

$0.04 per 1M input tokens

Fastino · checked 2026-11-02

Fastino docs: $0.04 input and $0 output for fastino/GLiDE.

No output-token charge does not mean free requests. These are provider tariffs, separate from benchmark workload costs.

374 recorded observations. Published results retain their source and setup; no Examenos evaluation was run. Missing results are not zero. How to read these comparisons →

Official / vendor · Fastino GLiDE v1.2 post (vendor measured)

GLiDE v1.2 · Official / vendor · Fastino GLiDE v1.2 post (vendor measured)
Evaluation	Result	Evidence
Fastino · Decision Index public suite, own complete run	71.5	Setup & source Fastino run of the official harness on the current public suite. GLiDE v1.2, own complete run of the official harness. Printed in the vendor's own table. Fastino's own complete runs of the official harness and its own held-out panels, not a leaderboard result. Jev comparison columns name Jev 1.13, which is unregistered, so they are not stored. Not ranked here. Checked / reported: 2026-11-02 · Samples: Not published Snapshot: 2026-10-22 · Revision: Read 2026-11-02 · Published observations; result reuse terms not specified Result source · Benchmark source
Fastino · Decision Index 0.2.1 edition, own run	71.12	Setup & source The edition of the launch post: 64.81 for the September checkpoint, and 71.12 for v1.2 on what Fastino calls the same edition. Own runs with the official scorer; Fastino prints v1.1 at 65.12, which is unregistered. Printed in the vendor's own table. Fastino's own complete runs of the official harness and its own held-out panels, not a leaderboard result. Jev comparison columns name Jev 1.13, which is unregistered, so they are not stored. Not ranked here. Checked / reported: 2026-11-02 · Samples: Not published Snapshot: 2026-10-22 · Revision: Read 2026-11-02 · Published observations; result reuse terms not specified Result source · Benchmark source
Fastino held-out panels · All eight panels · accuracy	83.0%	Setup & source All eight panels: 4,482 held-out items. One forward pass, no temperature tuning. Printed in the vendor's own table. Fastino's own complete runs of the official harness and its own held-out panels, not a leaderboard result. Jev comparison columns name Jev 1.13, which is unregistered, so they are not stored. Not ranked here. Checked / reported: 2026-11-02 · Samples: 4482 Snapshot: 2026-10-22 · Revision: Read 2026-11-02 · Published observations; result reuse terms not specified Result source · Benchmark source
Fastino held-out panels · All eight panels · calibration error	1.9 pp	Setup & source All eight panels: 4,482 held-out items; lower is better. One forward pass, no temperature tuning. Printed in the vendor's own table. Fastino's own complete runs of the official harness and its own held-out panels, not a leaderboard result. Jev comparison columns name Jev 1.13, which is unregistered, so they are not stored. Not ranked here. Checked / reported: 2026-11-02 · Samples: 4482 Snapshot: 2026-10-22 · Revision: Read 2026-11-02 · Published observations; result reuse terms not specified Result source · Benchmark source
Fastino held-out panels · Same-skill decisions · accuracy	90.2%	Setup & source Same-skill decisions: 1,282 held-out items. One forward pass, no temperature tuning. Printed in the vendor's own table. Fastino's own complete runs of the official harness and its own held-out panels, not a leaderboard result. Jev comparison columns name Jev 1.13, which is unregistered, so they are not stored. Not ranked here. Checked / reported: 2026-11-02 · Samples: 1282 Snapshot: 2026-10-22 · Revision: Read 2026-11-02 · Published observations; result reuse terms not specified Result source · Benchmark source
Fastino held-out panels · Same-skill decisions · calibration error	2.4 pp	Setup & source Same-skill decisions: 1,282 held-out items; lower is better. One forward pass, no temperature tuning. Printed in the vendor's own table. Fastino's own complete runs of the official harness and its own held-out panels, not a leaderboard result. Jev comparison columns name Jev 1.13, which is unregistered, so they are not stored. Not ranked here. Checked / reported: 2026-11-02 · Samples: 1282 Snapshot: 2026-10-22 · Revision: Read 2026-11-02 · Published observations; result reuse terms not specified Result source · Benchmark source
Fastino held-out panels · Independent new-domain decisions · accuracy	94.0%	Setup & source Independent new-domain decisions: 554 held-out items. One forward pass, no temperature tuning. Printed in the vendor's own table. Fastino's own complete runs of the official harness and its own held-out panels, not a leaderboard result. Jev comparison columns name Jev 1.13, which is unregistered, so they are not stored. Not ranked here. Checked / reported: 2026-11-02 · Samples: 554 Snapshot: 2026-10-22 · Revision: Read 2026-11-02 · Published observations; result reuse terms not specified Result source · Benchmark source
Fastino held-out panels · Independent new-domain decisions · calibration error	2.6 pp	Setup & source Independent new-domain decisions: 554 held-out items; lower is better. One forward pass, no temperature tuning. Printed in the vendor's own table. Fastino's own complete runs of the official harness and its own held-out panels, not a leaderboard result. Jev comparison columns name Jev 1.13, which is unregistered, so they are not stored. Not ranked here. Checked / reported: 2026-11-02 · Samples: 554 Snapshot: 2026-10-22 · Revision: Read 2026-11-02 · Published observations; result reuse terms not specified Result source · Benchmark source
Fastino held-out panels · Tool-call decisions · accuracy	98.3%	Setup & source Tool-call decisions: 300 held-out items. One forward pass, no temperature tuning. Printed in the vendor's own table. Fastino's own complete runs of the official harness and its own held-out panels, not a leaderboard result. Jev comparison columns name Jev 1.13, which is unregistered, so they are not stored. Not ranked here. Checked / reported: 2026-11-02 · Samples: 300 Snapshot: 2026-10-22 · Revision: Read 2026-11-02 · Published observations; result reuse terms not specified Result source · Benchmark source
Fastino held-out panels · Tool-call decisions · calibration error	1.8 pp	Setup & source Tool-call decisions: 300 held-out items; lower is better. One forward pass, no temperature tuning. Printed in the vendor's own table. Fastino's own complete runs of the official harness and its own held-out panels, not a leaderboard result. Jev comparison columns name Jev 1.13, which is unregistered, so they are not stored. Not ranked here. Checked / reported: 2026-11-02 · Samples: 300 Snapshot: 2026-10-22 · Revision: Read 2026-11-02 · Published observations; result reuse terms not specified Result source · Benchmark source
Fastino held-out panels · Safety classification · accuracy	90.7%	Setup & source Safety classification: 300 held-out items. One forward pass, no temperature tuning. Printed in the vendor's own table. Fastino's own complete runs of the official harness and its own held-out panels, not a leaderboard result. Jev comparison columns name Jev 1.13, which is unregistered, so they are not stored. Not ranked here. Checked / reported: 2026-11-02 · Samples: 300 Snapshot: 2026-10-22 · Revision: Read 2026-11-02 · Published observations; result reuse terms not specified Result source · Benchmark source
Fastino held-out panels · Safety classification · calibration error	5.0 pp	Setup & source Safety classification: 300 held-out items; lower is better. One forward pass, no temperature tuning. Printed in the vendor's own table. Fastino's own complete runs of the official harness and its own held-out panels, not a leaderboard result. Jev comparison columns name Jev 1.13, which is unregistered, so they are not stored. Not ranked here. Checked / reported: 2026-11-02 · Samples: 300 Snapshot: 2026-10-22 · Revision: Read 2026-11-02 · Published observations; result reuse terms not specified Result source · Benchmark source
Fastino held-out panels · Expert multiple choice · accuracy	61.2%	Setup & source Expert multiple choice: 577 held-out items. One forward pass, no temperature tuning. Printed in the vendor's own table. Fastino's own complete runs of the official harness and its own held-out panels, not a leaderboard result. Jev comparison columns name Jev 1.13, which is unregistered, so they are not stored. Not ranked here. Checked / reported: 2026-11-02 · Samples: 577 Snapshot: 2026-10-22 · Revision: Read 2026-11-02 · Published observations; result reuse terms not specified Result source · Benchmark source
Fastino held-out panels · Expert multiple choice · calibration error	6.9 pp	Setup & source Expert multiple choice: 577 held-out items; lower is better. One forward pass, no temperature tuning. Printed in the vendor's own table. Fastino's own complete runs of the official harness and its own held-out panels, not a leaderboard result. Jev comparison columns name Jev 1.13, which is unregistered, so they are not stored. Not ranked here. Checked / reported: 2026-11-02 · Samples: 577 Snapshot: 2026-10-22 · Revision: Read 2026-11-02 · Published observations; result reuse terms not specified Result source · Benchmark source
Fastino · Median latency	82 ms	Setup & source One forward pass per question. One RTX PRO 6000, 52 GB of GPU memory. Printed in the vendor's own table. Fastino's own complete runs of the official harness and its own held-out panels, not a leaderboard result. Jev comparison columns name Jev 1.13, which is unregistered, so they are not stored. Not ranked here. Checked / reported: 2026-11-02 · Samples: Not published Snapshot: 2026-10-22 · Revision: Read 2026-11-02 · Published observations; result reuse terms not specified Result source · Benchmark source
Fastino · P80 latency	245 ms	Setup & source One forward pass per question. One RTX PRO 6000, 52 GB of GPU memory. Printed in the vendor's own table. Fastino's own complete runs of the official harness and its own held-out panels, not a leaderboard result. Jev comparison columns name Jev 1.13, which is unregistered, so they are not stored. Not ranked here. Checked / reported: 2026-11-02 · Samples: Not published Snapshot: 2026-10-22 · Revision: Read 2026-11-02 · Published observations; result reuse terms not specified Result source · Benchmark source
Fastino · Median latency with images	93 ms	Setup & source A full request with images. One RTX PRO 6000, 52 GB of GPU memory. Printed in the vendor's own table. Fastino's own complete runs of the official harness and its own held-out panels, not a leaderboard result. Jev comparison columns name Jev 1.13, which is unregistered, so they are not stored. Not ranked here. Checked / reported: 2026-11-02 · Samples: Not published Snapshot: 2026-10-22 · Revision: Read 2026-11-02 · Published observations; result reuse terms not specified Result source · Benchmark source

Evaluator-published · Community Decision Index

Evaluator-published · Benchmark Heaven

GLiDE v1.2 · Evaluator-published · Benchmark Heaven
Evaluation	Result	Evidence
Benchmark Heaven JevBench v1.8 · Capability Score	84.923	Setup & source Capability Score on the rendered board. Fastino GLiDE 1.2 (open weights, self-hosted) · Open weights · LLM decoder. Board key fastino-glide-1.2-open; method tag v1.7. Measured on the Second regular draw (v1.7.0), 2026-10-23; put on the v1.6.1 scale by the draw's anchor offset (see How it works). Source link https://huggingface.co/fastino/GLiDE. Rendered board row. Benchmark Heaven's rendered board, read 2 November 2026: 283 ranked systems · 1,500 decisions each (600 for 19 equated API re-runs) · 131 newer ranked rows tagged with their draw · 14 older rows dated separately. Newer rows are measured on dated draws and put on the v1.6.1 scale by the draw's anchor offset, so a row's draw is part of its identity. Distinct from Leanmcp JevBench, the community Decision Index and the stored Benchmark Heaven v1.6.1 feed. Cost is the source's model on a GPU-time basis for self-hosted rows, not a provider tariff. Self-hosted latency is timed on the GPU over loopback and API latency end to end with the network; the source says they are not directly comparable, so they stay in separate datasets. No Examenos recomputation. Checked / reported: 2026-11-02 · Evaluated: 2026-10-23 · Samples: Not published Snapshot: 2026-10-31 · Revision: Board v1.7.48, method v1.8, updated 2026-10-31; page sha256 df9093384c9cdf5854e065f71dd3abd83a127793f22f4ca70268e9981e0a0b17 · Published summary observations; upstream task/result terms retained Result source · Benchmark source
Benchmark Heaven JevBench v1.8 · Intelligence	76.4	Setup & source Intelligence on the rendered board. Fastino GLiDE 1.2 (open weights, self-hosted) · Open weights · LLM decoder. Board key fastino-glide-1.2-open; method tag v1.7. Measured on the Second regular draw (v1.7.0), 2026-10-23; put on the v1.6.1 scale by the draw's anchor offset (see How it works). Source link https://huggingface.co/fastino/GLiDE. Rendered board row. Benchmark Heaven's rendered board, read 2 November 2026: 283 ranked systems · 1,500 decisions each (600 for 19 equated API re-runs) · 131 newer ranked rows tagged with their draw · 14 older rows dated separately. Newer rows are measured on dated draws and put on the v1.6.1 scale by the draw's anchor offset, so a row's draw is part of its identity. Distinct from Leanmcp JevBench, the community Decision Index and the stored Benchmark Heaven v1.6.1 feed. Cost is the source's model on a GPU-time basis for self-hosted rows, not a provider tariff. Self-hosted latency is timed on the GPU over loopback and API latency end to end with the network; the source says they are not directly comparable, so they stay in separate datasets. No Examenos recomputation. Checked / reported: 2026-11-02 · Evaluated: 2026-10-23 · Samples: Not published Snapshot: 2026-10-31 · Revision: Board v1.7.48, method v1.8, updated 2026-10-31; page sha256 df9093384c9cdf5854e065f71dd3abd83a127793f22f4ca70268e9981e0a0b17 · Published summary observations; upstream task/result terms retained Result source · Benchmark source
Benchmark Heaven JevBench v1.8 · Calibration	93.5	Setup & source Calibration on the rendered board. Fastino GLiDE 1.2 (open weights, self-hosted) · Open weights · LLM decoder. Board key fastino-glide-1.2-open; method tag v1.7. Measured on the Second regular draw (v1.7.0), 2026-10-23; put on the v1.6.1 scale by the draw's anchor offset (see How it works). Source link https://huggingface.co/fastino/GLiDE. Rendered board row. Benchmark Heaven's rendered board, read 2 November 2026: 283 ranked systems · 1,500 decisions each (600 for 19 equated API re-runs) · 131 newer ranked rows tagged with their draw · 14 older rows dated separately. Newer rows are measured on dated draws and put on the v1.6.1 scale by the draw's anchor offset, so a row's draw is part of its identity. Distinct from Leanmcp JevBench, the community Decision Index and the stored Benchmark Heaven v1.6.1 feed. Cost is the source's model on a GPU-time basis for self-hosted rows, not a provider tariff. Self-hosted latency is timed on the GPU over loopback and API latency end to end with the network; the source says they are not directly comparable, so they stay in separate datasets. No Examenos recomputation. Checked / reported: 2026-11-02 · Evaluated: 2026-10-23 · Samples: Not published Snapshot: 2026-10-31 · Revision: Board v1.7.48, method v1.8, updated 2026-10-31; page sha256 df9093384c9cdf5854e065f71dd3abd83a127793f22f4ca70268e9981e0a0b17 · Published summary observations; upstream task/result terms retained Result source · Benchmark source
Benchmark Heaven JevBench v1.8 · Cost Score	58.1	Setup & source Cost Score on the rendered board. Fastino GLiDE 1.2 (open weights, self-hosted) · Open weights · LLM decoder. Board key fastino-glide-1.2-open; method tag v1.7. Measured on the Second regular draw (v1.7.0), 2026-10-23; put on the v1.6.1 scale by the draw's anchor offset (see How it works). Source link https://huggingface.co/fastino/GLiDE. Rendered board row. Benchmark Heaven's rendered board, read 2 November 2026: 283 ranked systems · 1,500 decisions each (600 for 19 equated API re-runs) · 131 newer ranked rows tagged with their draw · 14 older rows dated separately. Newer rows are measured on dated draws and put on the v1.6.1 scale by the draw's anchor offset, so a row's draw is part of its identity. Distinct from Leanmcp JevBench, the community Decision Index and the stored Benchmark Heaven v1.6.1 feed. Cost is the source's model on a GPU-time basis for self-hosted rows, not a provider tariff. Self-hosted latency is timed on the GPU over loopback and API latency end to end with the network; the source says they are not directly comparable, so they stay in separate datasets. No Examenos recomputation. Checked / reported: 2026-11-02 · Evaluated: 2026-10-23 · Samples: Not published Snapshot: 2026-10-31 · Revision: Board v1.7.48, method v1.8, updated 2026-10-31; page sha256 df9093384c9cdf5854e065f71dd3abd83a127793f22f4ca70268e9981e0a0b17 · Published summary observations; upstream task/result terms retained Result source · Benchmark source
Benchmark Heaven JevBench v1.8 · Modelled cost per 1,000 decisions	0.02492 USD/1k decisions	Setup & source Modelled cost per 1,000 decisions on the rendered board. Fastino GLiDE 1.2 (open weights, self-hosted) · Open weights · LLM decoder. Board key fastino-glide-1.2-open; method tag v1.7. Measured on the Second regular draw (v1.7.0), 2026-10-23; put on the v1.6.1 scale by the draw's anchor offset (see How it works). Source link https://huggingface.co/fastino/GLiDE. Rendered board row. Benchmark Heaven's rendered board, read 2 November 2026: 283 ranked systems · 1,500 decisions each (600 for 19 equated API re-runs) · 131 newer ranked rows tagged with their draw · 14 older rows dated separately. Newer rows are measured on dated draws and put on the v1.6.1 scale by the draw's anchor offset, so a row's draw is part of its identity. Distinct from Leanmcp JevBench, the community Decision Index and the stored Benchmark Heaven v1.6.1 feed. Cost is the source's model on a GPU-time basis for self-hosted rows, not a provider tariff. Self-hosted latency is timed on the GPU over loopback and API latency end to end with the network; the source says they are not directly comparable, so they stay in separate datasets. No Examenos recomputation. Checked / reported: 2026-11-02 · Evaluated: 2026-10-23 · Samples: Not published Snapshot: 2026-10-31 · Revision: Board v1.7.48, method v1.8, updated 2026-10-31; page sha256 df9093384c9cdf5854e065f71dd3abd83a127793f22f4ca70268e9981e0a0b17 · Published summary observations; upstream task/result terms retained Result source · Benchmark source
Benchmark Heaven JevBench v1.8 · Self-hosted p50 on the GPU, loopback	0.140 s	Setup & source Self-hosted p50 on the GPU, loopback on the rendered board. Fastino GLiDE 1.2 (open weights, self-hosted) · Open weights · LLM decoder. Board key fastino-glide-1.2-open; method tag v1.7. Measured on the Second regular draw (v1.7.0), 2026-10-23; put on the v1.6.1 scale by the draw's anchor offset (see How it works). Source link https://huggingface.co/fastino/GLiDE. Rendered board row. Benchmark Heaven's rendered board, read 2 November 2026: 283 ranked systems · 1,500 decisions each (600 for 19 equated API re-runs) · 131 newer ranked rows tagged with their draw · 14 older rows dated separately. Newer rows are measured on dated draws and put on the v1.6.1 scale by the draw's anchor offset, so a row's draw is part of its identity. Distinct from Leanmcp JevBench, the community Decision Index and the stored Benchmark Heaven v1.6.1 feed. Cost is the source's model on a GPU-time basis for self-hosted rows, not a provider tariff. Self-hosted latency is timed on the GPU over loopback and API latency end to end with the network; the source says they are not directly comparable, so they stay in separate datasets. No Examenos recomputation. Checked / reported: 2026-11-02 · Evaluated: 2026-10-23 · Samples: Not published Snapshot: 2026-10-31 · Revision: Board v1.7.48, method v1.8, updated 2026-10-31; page sha256 df9093384c9cdf5854e065f71dd3abd83a127793f22f4ca70268e9981e0a0b17 · Published summary observations; upstream task/result terms retained Result source · Benchmark source
Benchmark Heaven JevBench v1.8 · Self-hosted estimate including network (2x + 150 ms)	0.43 s	Setup & source Self-hosted estimate including network (2x + 150 ms) on the rendered board. Fastino GLiDE 1.2 (open weights, self-hosted) · Open weights · LLM decoder. Board key fastino-glide-1.2-open; method tag v1.7. Measured on the Second regular draw (v1.7.0), 2026-10-23; put on the v1.6.1 scale by the draw's anchor offset (see How it works). Source link https://huggingface.co/fastino/GLiDE. Rendered board row. Benchmark Heaven's rendered board, read 2 November 2026: 283 ranked systems · 1,500 decisions each (600 for 19 equated API re-runs) · 131 newer ranked rows tagged with their draw · 14 older rows dated separately. Newer rows are measured on dated draws and put on the v1.6.1 scale by the draw's anchor offset, so a row's draw is part of its identity. Distinct from Leanmcp JevBench, the community Decision Index and the stored Benchmark Heaven v1.6.1 feed. Cost is the source's model on a GPU-time basis for self-hosted rows, not a provider tariff. Self-hosted latency is timed on the GPU over loopback and API latency end to end with the network; the source says they are not directly comparable, so they stay in separate datasets. No Examenos recomputation. Checked / reported: 2026-11-02 · Evaluated: 2026-10-23 · Samples: Not published Snapshot: 2026-10-31 · Revision: Board v1.7.48, method v1.8, updated 2026-10-31; page sha256 df9093384c9cdf5854e065f71dd3abd83a127793f22f4ca70268e9981e0a0b17 · Published summary observations; upstream task/result terms retained Result source · Benchmark source

GLiDE v1.2 · Evaluator-published · Community Decision Index
Evaluation	Result	Evidence
HF text 0.4.1 · Full score	71.52	Setup & source HF text 0.4.1 · Full score Exact source engine glide-v1.2; {"kind": "full fine-tune", "technique": "autoregressive", "technique_name": "autoregressive", "served_params": 27800000000, "params_estimated": false, "params_from_name": false, "trained_bytes": null, "model_url": "https://huggingface.co/fastino/GLiDE", "code_url": "https://huggingface.co/fastino/GLiDE/tree/main/glide", "base_model": "Qwen/Qwen3.8-27B", "base_url": "https://huggingface.co/Qwen/Qwen3.8-27B", "weights_repo": "fastino/GLiDE", "served_checkpoint": "a7e45896be7c96ba3354891da790c5ec99c30b6c", "weights_note": null, "org": {"id": "fastino", "name": "Fastino", "logo": "data/orgs/fastino.png"}, "pr": 255, "replaces": "decision-27b-ckpt-nothink", "new": true}. New in the 2026-10-21 PR batch (PR #255). Replaces the earlier board entry decision-27b-ckpt-nothink (newer version of the same model). Submitted public 0.3 run adopted (kit rescore 71.99). Private tests run by the lab on one RTX PRO 6000 with the author's serving; decision agreement with the submitted rows 99.90% on 895 rows (bar 98.5%). Weights are public on the Hugging Face Hub (fastino-community license); the lab scored revision a7e4589. Decision Index 0.4.1 (release-v4), source generated 2026-10-29T14:04:16Z. Full = 60% equated private same-skill + 40% equated private new-domain tasks; public contributes zero. Both private components are mapped onto the public scale using untuned stock models. Public weighting/panel changed (GSM8K rebuilt, ForecastBench retired, WinoGrande area reassigned); these are new benchmark definitions, not proof of a model upgrade. Public 37-benchmark index and 42 published native benchmarks remain separate. Preserve source rounding, coverage, exclusions and unknown execution dates; no new uncertainty inferred. Source notes: New in the 2026-10-21 PR batch (PR #255). Replaces the earlier board entry decision-27b-ckpt-nothink (newer version of the same model). Submitted public 0.3 run adopted (kit rescore 71.99). Private tests run by the lab on one RTX PRO 6000 with the author's serving; decision agreement with the submitted rows 99.90% on 895 rows (bar 98.5%). Weights are public on the Hugging Face Hub (fastino-community license); the lab scored revision a7e4589. Author public-run adoption/execution status and private lab setup are disclosed in configuration; evaluator-published does not mean every public run was independently executed. The board served a7e45896be; the registry pins the newest repository revision, and the commit titles after it concern the card. Community Decision Index 0.4.1: expanded/refreshed harder private new-domain tasks and equating; Full remains 60% private same-skill +40% private new domains, zero public contribution. This population revision is not a model upgrade/regression. Public/native/calibration data remain on their unchanged 0.4 definitions. Source publication/generation is not a model run date. Public submitted runs plus evaluator private measurements are Evaluator-published, not wholly independently executed. Checked / reported: 2026-11-02 · Samples: Not published Snapshot: 2026-10-22T05:40:06Z · Revision: d9962113a1acc95c2d7522c2babfb989afbeb034; c130faf65f82e81d3bdefb01e758c712dc5c5199 for the 2 November 2026 additions · Published observations; result reuse terms not specified Result source · Benchmark source
HF text 0.4 · Public index	71.99	Setup & source HF text 0.4 · Public index Exact source engine glide-v1.2; {"kind": "full fine-tune", "technique": "autoregressive", "technique_name": "autoregressive", "served_params": 27800000000, "params_estimated": false, "params_from_name": false, "trained_bytes": null, "model_url": "https://huggingface.co/fastino/GLiDE", "code_url": "https://huggingface.co/fastino/GLiDE/tree/main/glide", "base_model": "Qwen/Qwen3.8-27B", "base_url": "https://huggingface.co/Qwen/Qwen3.8-27B", "weights_repo": "fastino/GLiDE", "served_checkpoint": "a7e45896be7c96ba3354891da790c5ec99c30b6c", "weights_note": null, "org": {"id": "fastino", "name": "Fastino", "logo": "data/orgs/fastino.png"}, "pr": 255, "replaces": "decision-27b-ckpt-nothink", "new": true}. New in the 2026-10-21 PR batch (PR #255). Replaces the earlier board entry decision-27b-ckpt-nothink (newer version of the same model). Submitted public 0.3 run adopted (kit rescore 71.99). Private tests run by the lab on one RTX PRO 6000 with the author's serving; decision agreement with the submitted rows 99.90% on 895 rows (bar 98.5%). Weights are public on the Hugging Face Hub (fastino-community license); the lab scored revision a7e4589. Decision Index 0.4.1 (release-v4), source generated 2026-10-29T14:04:16Z. Full = 60% equated private same-skill + 40% equated private new-domain tasks; public contributes zero. Both private components are mapped onto the public scale using untuned stock models. Public weighting/panel changed (GSM8K rebuilt, ForecastBench retired, WinoGrande area reassigned); these are new benchmark definitions, not proof of a model upgrade. Public 37-benchmark index and 42 published native benchmarks remain separate. Preserve source rounding, coverage, exclusions and unknown execution dates; no new uncertainty inferred. Source notes: New in the 2026-10-21 PR batch (PR #255). Replaces the earlier board entry decision-27b-ckpt-nothink (newer version of the same model). Submitted public 0.3 run adopted (kit rescore 71.99). Private tests run by the lab on one RTX PRO 6000 with the author's serving; decision agreement with the submitted rows 99.90% on 895 rows (bar 98.5%). Weights are public on the Hugging Face Hub (fastino-community license); the lab scored revision a7e4589. Author public-run adoption/execution status and private lab setup are disclosed in configuration; evaluator-published does not mean every public run was independently executed. The board served a7e45896be; the registry pins the newest repository revision, and the commit titles after it concern the card. Decision Index 0.4 (release-v4), source generated 2026-10-20T14:13:56Z. Full = 60% equated private same-skill + 40% equated private new-domain tasks; public contributes zero. Both private components are mapped onto the public scale using untuned stock models. Public weighting/panel changed (GSM8K rebuilt, ForecastBench retired, WinoGrande area reassigned); these are new benchmark definitions, not proof of a model upgrade. Public 37-benchmark index and 42 published native benchmarks remain separate. Preserve source rounding, coverage, exclusions and unknown execution dates; no new uncertainty inferred. Checked / reported: 2026-11-02 · Samples: Not published Snapshot: 2026-10-20T14:13:56Z · Revision: 133313cf8963f2b6bed960abb0152fd14cbb6731; c130faf65f82e81d3bdefb01e758c712dc5c5199 for the 2 November 2026 additions · Published observations; result reuse terms not specified Result source · Benchmark source
HF text 0.4 · Private same-skill component	68.57	Setup & source HF text 0.4 · Private same-skill component Exact source engine glide-v1.2; {"kind": "full fine-tune", "technique": "autoregressive", "technique_name": "autoregressive", "served_params": 27800000000, "params_estimated": false, "params_from_name": false, "trained_bytes": null, "model_url": "https://huggingface.co/fastino/GLiDE", "code_url": "https://huggingface.co/fastino/GLiDE/tree/main/glide", "base_model": "Qwen/Qwen3.8-27B", "base_url": "https://huggingface.co/Qwen/Qwen3.8-27B", "weights_repo": "fastino/GLiDE", "served_checkpoint": "a7e45896be7c96ba3354891da790c5ec99c30b6c", "weights_note": null, "org": {"id": "fastino", "name": "Fastino", "logo": "data/orgs/fastino.png"}, "pr": 255, "replaces": "decision-27b-ckpt-nothink", "new": true}. New in the 2026-10-21 PR batch (PR #255). Replaces the earlier board entry decision-27b-ckpt-nothink (newer version of the same model). Submitted public 0.3 run adopted (kit rescore 71.99). Private tests run by the lab on one RTX PRO 6000 with the author's serving; decision agreement with the submitted rows 99.90% on 895 rows (bar 98.5%). Weights are public on the Hugging Face Hub (fastino-community license); the lab scored revision a7e4589. Decision Index 0.4.1 (release-v4), source generated 2026-10-29T14:04:16Z. Full = 60% equated private same-skill + 40% equated private new-domain tasks; public contributes zero. Both private components are mapped onto the public scale using untuned stock models. Public weighting/panel changed (GSM8K rebuilt, ForecastBench retired, WinoGrande area reassigned); these are new benchmark definitions, not proof of a model upgrade. Public 37-benchmark index and 42 published native benchmarks remain separate. Preserve source rounding, coverage, exclusions and unknown execution dates; no new uncertainty inferred. Source notes: New in the 2026-10-21 PR batch (PR #255). Replaces the earlier board entry decision-27b-ckpt-nothink (newer version of the same model). Submitted public 0.3 run adopted (kit rescore 71.99). Private tests run by the lab on one RTX PRO 6000 with the author's serving; decision agreement with the submitted rows 99.90% on 895 rows (bar 98.5%). Weights are public on the Hugging Face Hub (fastino-community license); the lab scored revision a7e4589. Author public-run adoption/execution status and private lab setup are disclosed in configuration; evaluator-published does not mean every public run was independently executed. The board served a7e45896be; the registry pins the newest repository revision, and the commit titles after it concern the card. Decision Index 0.4 (release-v4), source generated 2026-10-20T14:13:56Z. Full = 60% equated private same-skill + 40% equated private new-domain tasks; public contributes zero. Both private components are mapped onto the public scale using untuned stock models. Public weighting/panel changed (GSM8K rebuilt, ForecastBench retired, WinoGrande area reassigned); these are new benchmark definitions, not proof of a model upgrade. Public 37-benchmark index and 42 published native benchmarks remain separate. Preserve source rounding, coverage, exclusions and unknown execution dates; no new uncertainty inferred. Checked / reported: 2026-11-02 · Samples: Not published Snapshot: 2026-10-20T14:13:56Z · Revision: 133313cf8963f2b6bed960abb0152fd14cbb6731; c130faf65f82e81d3bdefb01e758c712dc5c5199 for the 2 November 2026 additions · Published observations; result reuse terms not specified Result source · Benchmark source
HF text 0.4.1 · Private new-domain component	59.66	Setup & source HF text 0.4.1 · Private new-domain component Exact source engine glide-v1.2; {"kind": "full fine-tune", "technique": "autoregressive", "technique_name": "autoregressive", "served_params": 27800000000, "params_estimated": false, "params_from_name": false, "trained_bytes": null, "model_url": "https://huggingface.co/fastino/GLiDE", "code_url": "https://huggingface.co/fastino/GLiDE/tree/main/glide", "base_model": "Qwen/Qwen3.8-27B", "base_url": "https://huggingface.co/Qwen/Qwen3.8-27B", "weights_repo": "fastino/GLiDE", "served_checkpoint": "a7e45896be7c96ba3354891da790c5ec99c30b6c", "weights_note": null, "org": {"id": "fastino", "name": "Fastino", "logo": "data/orgs/fastino.png"}, "pr": 255, "replaces": "decision-27b-ckpt-nothink", "new": true}. New in the 2026-10-21 PR batch (PR #255). Replaces the earlier board entry decision-27b-ckpt-nothink (newer version of the same model). Submitted public 0.3 run adopted (kit rescore 71.99). Private tests run by the lab on one RTX PRO 6000 with the author's serving; decision agreement with the submitted rows 99.90% on 895 rows (bar 98.5%). Weights are public on the Hugging Face Hub (fastino-community license); the lab scored revision a7e4589. Decision Index 0.4.1 (release-v4), source generated 2026-10-29T14:04:16Z. Full = 60% equated private same-skill + 40% equated private new-domain tasks; public contributes zero. Both private components are mapped onto the public scale using untuned stock models. Public weighting/panel changed (GSM8K rebuilt, ForecastBench retired, WinoGrande area reassigned); these are new benchmark definitions, not proof of a model upgrade. Public 37-benchmark index and 42 published native benchmarks remain separate. Preserve source rounding, coverage, exclusions and unknown execution dates; no new uncertainty inferred. Source notes: New in the 2026-10-21 PR batch (PR #255). Replaces the earlier board entry decision-27b-ckpt-nothink (newer version of the same model). Submitted public 0.3 run adopted (kit rescore 71.99). Private tests run by the lab on one RTX PRO 6000 with the author's serving; decision agreement with the submitted rows 99.90% on 895 rows (bar 98.5%). Weights are public on the Hugging Face Hub (fastino-community license); the lab scored revision a7e4589. Author public-run adoption/execution status and private lab setup are disclosed in configuration; evaluator-published does not mean every public run was independently executed. The board served a7e45896be; the registry pins the newest repository revision, and the commit titles after it concern the card. Community Decision Index 0.4.1: expanded/refreshed harder private new-domain tasks and equating; Full remains 60% private same-skill +40% private new domains, zero public contribution. This population revision is not a model upgrade/regression. Public/native/calibration data remain on their unchanged 0.4 definitions. Source publication/generation is not a model run date. Public submitted runs plus evaluator private measurements are Evaluator-published, not wholly independently executed. Checked / reported: 2026-11-02 · Samples: Not published Snapshot: 2026-10-22T05:40:06Z · Revision: d9962113a1acc95c2d7522c2babfb989afbeb034; c130faf65f82e81d3bdefb01e758c712dc5c5199 for the 2 November 2026 additions · Published observations; result reuse terms not specified Result source · Benchmark source
HF text 0.4 · Raw public index	78.89	Setup & source HF text 0.4 · Raw public index Exact source engine glide-v1.2; {"kind": "full fine-tune", "technique": "autoregressive", "technique_name": "autoregressive", "served_params": 27800000000, "params_estimated": false, "params_from_name": false, "trained_bytes": null, "model_url": "https://huggingface.co/fastino/GLiDE", "code_url": "https://huggingface.co/fastino/GLiDE/tree/main/glide", "base_model": "Qwen/Qwen3.8-27B", "base_url": "https://huggingface.co/Qwen/Qwen3.8-27B", "weights_repo": "fastino/GLiDE", "served_checkpoint": "a7e45896be7c96ba3354891da790c5ec99c30b6c", "weights_note": null, "org": {"id": "fastino", "name": "Fastino", "logo": "data/orgs/fastino.png"}, "pr": 255, "replaces": "decision-27b-ckpt-nothink", "new": true}. New in the 2026-10-21 PR batch (PR #255). Replaces the earlier board entry decision-27b-ckpt-nothink (newer version of the same model). Submitted public 0.3 run adopted (kit rescore 71.99). Private tests run by the lab on one RTX PRO 6000 with the author's serving; decision agreement with the submitted rows 99.90% on 895 rows (bar 98.5%). Weights are public on the Hugging Face Hub (fastino-community license); the lab scored revision a7e4589. Decision Index 0.4.1 (release-v4), source generated 2026-10-29T14:04:16Z. Full = 60% equated private same-skill + 40% equated private new-domain tasks; public contributes zero. Both private components are mapped onto the public scale using untuned stock models. Public weighting/panel changed (GSM8K rebuilt, ForecastBench retired, WinoGrande area reassigned); these are new benchmark definitions, not proof of a model upgrade. Public 37-benchmark index and 42 published native benchmarks remain separate. Preserve source rounding, coverage, exclusions and unknown execution dates; no new uncertainty inferred. Source notes: New in the 2026-10-21 PR batch (PR #255). Replaces the earlier board entry decision-27b-ckpt-nothink (newer version of the same model). Submitted public 0.3 run adopted (kit rescore 71.99). Private tests run by the lab on one RTX PRO 6000 with the author's serving; decision agreement with the submitted rows 99.90% on 895 rows (bar 98.5%). Weights are public on the Hugging Face Hub (fastino-community license); the lab scored revision a7e4589. Author public-run adoption/execution status and private lab setup are disclosed in configuration; evaluator-published does not mean every public run was independently executed. The board served a7e45896be; the registry pins the newest repository revision, and the commit titles after it concern the card. Decision Index 0.4 (release-v4), source generated 2026-10-20T14:13:56Z. Full = 60% equated private same-skill + 40% equated private new-domain tasks; public contributes zero. Both private components are mapped onto the public scale using untuned stock models. Public weighting/panel changed (GSM8K rebuilt, ForecastBench retired, WinoGrande area reassigned); these are new benchmark definitions, not proof of a model upgrade. Public 37-benchmark index and 42 published native benchmarks remain separate. Preserve source rounding, coverage, exclusions and unknown execution dates; no new uncertainty inferred. Checked / reported: 2026-11-02 · Samples: Not published Snapshot: 2026-10-20T14:13:56Z · Revision: 133313cf8963f2b6bed960abb0152fd14cbb6731; c130faf65f82e81d3bdefb01e758c712dc5c5199 for the 2 November 2026 additions · Published observations; result reuse terms not specified Result source · Benchmark source
HF text 0.4 · Public breadth index	71.44	Setup & source HF text 0.4 · Public breadth index Exact source engine glide-v1.2; {"kind": "full fine-tune", "technique": "autoregressive", "technique_name": "autoregressive", "served_params": 27800000000, "params_estimated": false, "params_from_name": false, "trained_bytes": null, "model_url": "https://huggingface.co/fastino/GLiDE", "code_url": "https://huggingface.co/fastino/GLiDE/tree/main/glide", "base_model": "Qwen/Qwen3.8-27B", "base_url": "https://huggingface.co/Qwen/Qwen3.8-27B", "weights_repo": "fastino/GLiDE", "served_checkpoint": "a7e45896be7c96ba3354891da790c5ec99c30b6c", "weights_note": null, "org": {"id": "fastino", "name": "Fastino", "logo": "data/orgs/fastino.png"}, "pr": 255, "replaces": "decision-27b-ckpt-nothink", "new": true}. New in the 2026-10-21 PR batch (PR #255). Replaces the earlier board entry decision-27b-ckpt-nothink (newer version of the same model). Submitted public 0.3 run adopted (kit rescore 71.99). Private tests run by the lab on one RTX PRO 6000 with the author's serving; decision agreement with the submitted rows 99.90% on 895 rows (bar 98.5%). Weights are public on the Hugging Face Hub (fastino-community license); the lab scored revision a7e4589. Decision Index 0.4.1 (release-v4), source generated 2026-10-29T14:04:16Z. Full = 60% equated private same-skill + 40% equated private new-domain tasks; public contributes zero. Both private components are mapped onto the public scale using untuned stock models. Public weighting/panel changed (GSM8K rebuilt, ForecastBench retired, WinoGrande area reassigned); these are new benchmark definitions, not proof of a model upgrade. Public 37-benchmark index and 42 published native benchmarks remain separate. Preserve source rounding, coverage, exclusions and unknown execution dates; no new uncertainty inferred. Source notes: New in the 2026-10-21 PR batch (PR #255). Replaces the earlier board entry decision-27b-ckpt-nothink (newer version of the same model). Submitted public 0.3 run adopted (kit rescore 71.99). Private tests run by the lab on one RTX PRO 6000 with the author's serving; decision agreement with the submitted rows 99.90% on 895 rows (bar 98.5%). Weights are public on the Hugging Face Hub (fastino-community license); the lab scored revision a7e4589. Author public-run adoption/execution status and private lab setup are disclosed in configuration; evaluator-published does not mean every public run was independently executed. The board served a7e45896be; the registry pins the newest repository revision, and the commit titles after it concern the card. Decision Index 0.4 (release-v4), source generated 2026-10-20T14:13:56Z. Full = 60% equated private same-skill + 40% equated private new-domain tasks; public contributes zero. Both private components are mapped onto the public scale using untuned stock models. Public weighting/panel changed (GSM8K rebuilt, ForecastBench retired, WinoGrande area reassigned); these are new benchmark definitions, not proof of a model upgrade. Public 37-benchmark index and 42 published native benchmarks remain separate. Preserve source rounding, coverage, exclusions and unknown execution dates; no new uncertainty inferred. Checked / reported: 2026-11-02 · Samples: Not published Snapshot: 2026-10-20T14:13:56Z · Revision: 133313cf8963f2b6bed960abb0152fd14cbb6731; c130faf65f82e81d3bdefb01e758c712dc5c5199 for the 2 November 2026 additions · Published observations; result reuse terms not specified Result source · Benchmark source
HF text 0.4 · Equated private same-skill	68.31	Setup & source HF text 0.4 · Equated private same-skill Exact source engine glide-v1.2; {"kind": "full fine-tune", "technique": "autoregressive", "technique_name": "autoregressive", "served_params": 27800000000, "params_estimated": false, "params_from_name": false, "trained_bytes": null, "model_url": "https://huggingface.co/fastino/GLiDE", "code_url": "https://huggingface.co/fastino/GLiDE/tree/main/glide", "base_model": "Qwen/Qwen3.8-27B", "base_url": "https://huggingface.co/Qwen/Qwen3.8-27B", "weights_repo": "fastino/GLiDE", "served_checkpoint": "a7e45896be7c96ba3354891da790c5ec99c30b6c", "weights_note": null, "org": {"id": "fastino", "name": "Fastino", "logo": "data/orgs/fastino.png"}, "pr": 255, "replaces": "decision-27b-ckpt-nothink", "new": true}. New in the 2026-10-21 PR batch (PR #255). Replaces the earlier board entry decision-27b-ckpt-nothink (newer version of the same model). Submitted public 0.3 run adopted (kit rescore 71.99). Private tests run by the lab on one RTX PRO 6000 with the author's serving; decision agreement with the submitted rows 99.90% on 895 rows (bar 98.5%). Weights are public on the Hugging Face Hub (fastino-community license); the lab scored revision a7e4589. Decision Index 0.4.1 (release-v4), source generated 2026-10-29T14:04:16Z. Full = 60% equated private same-skill + 40% equated private new-domain tasks; public contributes zero. Both private components are mapped onto the public scale using untuned stock models. Public weighting/panel changed (GSM8K rebuilt, ForecastBench retired, WinoGrande area reassigned); these are new benchmark definitions, not proof of a model upgrade. Public 37-benchmark index and 42 published native benchmarks remain separate. Preserve source rounding, coverage, exclusions and unknown execution dates; no new uncertainty inferred. Source notes: New in the 2026-10-21 PR batch (PR #255). Replaces the earlier board entry decision-27b-ckpt-nothink (newer version of the same model). Submitted public 0.3 run adopted (kit rescore 71.99). Private tests run by the lab on one RTX PRO 6000 with the author's serving; decision agreement with the submitted rows 99.90% on 895 rows (bar 98.5%). Weights are public on the Hugging Face Hub (fastino-community license); the lab scored revision a7e4589. Author public-run adoption/execution status and private lab setup are disclosed in configuration; evaluator-published does not mean every public run was independently executed. The board served a7e45896be; the registry pins the newest repository revision, and the commit titles after it concern the card. Decision Index 0.4 (release-v4), source generated 2026-10-20T14:13:56Z. Full = 60% equated private same-skill + 40% equated private new-domain tasks; public contributes zero. Both private components are mapped onto the public scale using untuned stock models. Public weighting/panel changed (GSM8K rebuilt, ForecastBench retired, WinoGrande area reassigned); these are new benchmark definitions, not proof of a model upgrade. Public 37-benchmark index and 42 published native benchmarks remain separate. Preserve source rounding, coverage, exclusions and unknown execution dates; no new uncertainty inferred. Checked / reported: 2026-11-02 · Samples: Not published Snapshot: 2026-10-20T14:13:56Z · Revision: 133313cf8963f2b6bed960abb0152fd14cbb6731; c130faf65f82e81d3bdefb01e758c712dc5c5199 for the 2 November 2026 additions · Published observations; result reuse terms not specified Result source · Benchmark source
HF text 0.4.1 · Equated private new domains	76.32	Setup & source HF text 0.4.1 · Equated private new domains Exact source engine glide-v1.2; {"kind": "full fine-tune", "technique": "autoregressive", "technique_name": "autoregressive", "served_params": 27800000000, "params_estimated": false, "params_from_name": false, "trained_bytes": null, "model_url": "https://huggingface.co/fastino/GLiDE", "code_url": "https://huggingface.co/fastino/GLiDE/tree/main/glide", "base_model": "Qwen/Qwen3.8-27B", "base_url": "https://huggingface.co/Qwen/Qwen3.8-27B", "weights_repo": "fastino/GLiDE", "served_checkpoint": "a7e45896be7c96ba3354891da790c5ec99c30b6c", "weights_note": null, "org": {"id": "fastino", "name": "Fastino", "logo": "data/orgs/fastino.png"}, "pr": 255, "replaces": "decision-27b-ckpt-nothink", "new": true}. New in the 2026-10-21 PR batch (PR #255). Replaces the earlier board entry decision-27b-ckpt-nothink (newer version of the same model). Submitted public 0.3 run adopted (kit rescore 71.99). Private tests run by the lab on one RTX PRO 6000 with the author's serving; decision agreement with the submitted rows 99.90% on 895 rows (bar 98.5%). Weights are public on the Hugging Face Hub (fastino-community license); the lab scored revision a7e4589. Decision Index 0.4.1 (release-v4), source generated 2026-10-29T14:04:16Z. Full = 60% equated private same-skill + 40% equated private new-domain tasks; public contributes zero. Both private components are mapped onto the public scale using untuned stock models. Public weighting/panel changed (GSM8K rebuilt, ForecastBench retired, WinoGrande area reassigned); these are new benchmark definitions, not proof of a model upgrade. Public 37-benchmark index and 42 published native benchmarks remain separate. Preserve source rounding, coverage, exclusions and unknown execution dates; no new uncertainty inferred. Source notes: New in the 2026-10-21 PR batch (PR #255). Replaces the earlier board entry decision-27b-ckpt-nothink (newer version of the same model). Submitted public 0.3 run adopted (kit rescore 71.99). Private tests run by the lab on one RTX PRO 6000 with the author's serving; decision agreement with the submitted rows 99.90% on 895 rows (bar 98.5%). Weights are public on the Hugging Face Hub (fastino-community license); the lab scored revision a7e4589. Author public-run adoption/execution status and private lab setup are disclosed in configuration; evaluator-published does not mean every public run was independently executed. The board served a7e45896be; the registry pins the newest repository revision, and the commit titles after it concern the card. Community Decision Index 0.4.1: expanded/refreshed harder private new-domain tasks and equating; Full remains 60% private same-skill +40% private new domains, zero public contribution. This population revision is not a model upgrade/regression. Public/native/calibration data remain on their unchanged 0.4 definitions. Source publication/generation is not a model run date. Public submitted runs plus evaluator private measurements are Evaluator-published, not wholly independently executed. Checked / reported: 2026-11-02 · Samples: Not published Snapshot: 2026-10-22T05:40:06Z · Revision: d9962113a1acc95c2d7522c2babfb989afbeb034; c130faf65f82e81d3bdefb01e758c712dc5c5199 for the 2 November 2026 additions · Published observations; result reuse terms not specified Result source · Benchmark source
HF text 0.4 · Full contribution · public	0.00	Setup & source HF text 0.4 · Full contribution · public Exact source engine glide-v1.2; {"kind": "full fine-tune", "technique": "autoregressive", "technique_name": "autoregressive", "served_params": 27800000000, "params_estimated": false, "params_from_name": false, "trained_bytes": null, "model_url": "https://huggingface.co/fastino/GLiDE", "code_url": "https://huggingface.co/fastino/GLiDE/tree/main/glide", "base_model": "Qwen/Qwen3.8-27B", "base_url": "https://huggingface.co/Qwen/Qwen3.8-27B", "weights_repo": "fastino/GLiDE", "served_checkpoint": "a7e45896be7c96ba3354891da790c5ec99c30b6c", "weights_note": null, "org": {"id": "fastino", "name": "Fastino", "logo": "data/orgs/fastino.png"}, "pr": 255, "replaces": "decision-27b-ckpt-nothink", "new": true}. New in the 2026-10-21 PR batch (PR #255). Replaces the earlier board entry decision-27b-ckpt-nothink (newer version of the same model). Submitted public 0.3 run adopted (kit rescore 71.99). Private tests run by the lab on one RTX PRO 6000 with the author's serving; decision agreement with the submitted rows 99.90% on 895 rows (bar 98.5%). Weights are public on the Hugging Face Hub (fastino-community license); the lab scored revision a7e4589. Decision Index 0.4.1 (release-v4), source generated 2026-10-29T14:04:16Z. Full = 60% equated private same-skill + 40% equated private new-domain tasks; public contributes zero. Both private components are mapped onto the public scale using untuned stock models. Public weighting/panel changed (GSM8K rebuilt, ForecastBench retired, WinoGrande area reassigned); these are new benchmark definitions, not proof of a model upgrade. Public 37-benchmark index and 42 published native benchmarks remain separate. Preserve source rounding, coverage, exclusions and unknown execution dates; no new uncertainty inferred. Source notes: New in the 2026-10-21 PR batch (PR #255). Replaces the earlier board entry decision-27b-ckpt-nothink (newer version of the same model). Submitted public 0.3 run adopted (kit rescore 71.99). Private tests run by the lab on one RTX PRO 6000 with the author's serving; decision agreement with the submitted rows 99.90% on 895 rows (bar 98.5%). Weights are public on the Hugging Face Hub (fastino-community license); the lab scored revision a7e4589. Author public-run adoption/execution status and private lab setup are disclosed in configuration; evaluator-published does not mean every public run was independently executed. Published weighted contribution; not a distinct benchmark execution. The board served a7e45896be; the registry pins the newest repository revision, and the commit titles after it concern the card. Decision Index 0.4 (release-v4), source generated 2026-10-20T14:13:56Z. Full = 60% equated private same-skill + 40% equated private new-domain tasks; public contributes zero. Both private components are mapped onto the public scale using untuned stock models. Public weighting/panel changed (GSM8K rebuilt, ForecastBench retired, WinoGrande area reassigned); these are new benchmark definitions, not proof of a model upgrade. Public 37-benchmark index and 42 published native benchmarks remain separate. Preserve source rounding, coverage, exclusions and unknown execution dates; no new uncertainty inferred. Checked / reported: 2026-11-02 · Samples: Not published Snapshot: 2026-10-20T14:13:56Z · Revision: 133313cf8963f2b6bed960abb0152fd14cbb6731; c130faf65f82e81d3bdefb01e758c712dc5c5199 for the 2 November 2026 additions · Published observations; result reuse terms not specified Result source · Benchmark source
HF text 0.4 · Full contribution · same_skills	40.99	Setup & source HF text 0.4 · Full contribution · same_skills Exact source engine glide-v1.2; {"kind": "full fine-tune", "technique": "autoregressive", "technique_name": "autoregressive", "served_params": 27800000000, "params_estimated": false, "params_from_name": false, "trained_bytes": null, "model_url": "https://huggingface.co/fastino/GLiDE", "code_url": "https://huggingface.co/fastino/GLiDE/tree/main/glide", "base_model": "Qwen/Qwen3.8-27B", "base_url": "https://huggingface.co/Qwen/Qwen3.8-27B", "weights_repo": "fastino/GLiDE", "served_checkpoint": "a7e45896be7c96ba3354891da790c5ec99c30b6c", "weights_note": null, "org": {"id": "fastino", "name": "Fastino", "logo": "data/orgs/fastino.png"}, "pr": 255, "replaces": "decision-27b-ckpt-nothink", "new": true}. New in the 2026-10-21 PR batch (PR #255). Replaces the earlier board entry decision-27b-ckpt-nothink (newer version of the same model). Submitted public 0.3 run adopted (kit rescore 71.99). Private tests run by the lab on one RTX PRO 6000 with the author's serving; decision agreement with the submitted rows 99.90% on 895 rows (bar 98.5%). Weights are public on the Hugging Face Hub (fastino-community license); the lab scored revision a7e4589. Decision Index 0.4.1 (release-v4), source generated 2026-10-29T14:04:16Z. Full = 60% equated private same-skill + 40% equated private new-domain tasks; public contributes zero. Both private components are mapped onto the public scale using untuned stock models. Public weighting/panel changed (GSM8K rebuilt, ForecastBench retired, WinoGrande area reassigned); these are new benchmark definitions, not proof of a model upgrade. Public 37-benchmark index and 42 published native benchmarks remain separate. Preserve source rounding, coverage, exclusions and unknown execution dates; no new uncertainty inferred. Source notes: New in the 2026-10-21 PR batch (PR #255). Replaces the earlier board entry decision-27b-ckpt-nothink (newer version of the same model). Submitted public 0.3 run adopted (kit rescore 71.99). Private tests run by the lab on one RTX PRO 6000 with the author's serving; decision agreement with the submitted rows 99.90% on 895 rows (bar 98.5%). Weights are public on the Hugging Face Hub (fastino-community license); the lab scored revision a7e4589. Author public-run adoption/execution status and private lab setup are disclosed in configuration; evaluator-published does not mean every public run was independently executed. Published weighted contribution; not a distinct benchmark execution. The board served a7e45896be; the registry pins the newest repository revision, and the commit titles after it concern the card. Decision Index 0.4 (release-v4), source generated 2026-10-20T14:13:56Z. Full = 60% equated private same-skill + 40% equated private new-domain tasks; public contributes zero. Both private components are mapped onto the public scale using untuned stock models. Public weighting/panel changed (GSM8K rebuilt, ForecastBench retired, WinoGrande area reassigned); these are new benchmark definitions, not proof of a model upgrade. Public 37-benchmark index and 42 published native benchmarks remain separate. Preserve source rounding, coverage, exclusions and unknown execution dates; no new uncertainty inferred. Checked / reported: 2026-11-02 · Samples: Not published Snapshot: 2026-10-20T14:13:56Z · Revision: 133313cf8963f2b6bed960abb0152fd14cbb6731; c130faf65f82e81d3bdefb01e758c712dc5c5199 for the 2 November 2026 additions · Published observations; result reuse terms not specified Result source · Benchmark source
HF text 0.4.1 · Full contribution · new_domains	30.53	Setup & source HF text 0.4.1 · Full contribution · new_domains Exact source engine glide-v1.2; {"kind": "full fine-tune", "technique": "autoregressive", "technique_name": "autoregressive", "served_params": 27800000000, "params_estimated": false, "params_from_name": false, "trained_bytes": null, "model_url": "https://huggingface.co/fastino/GLiDE", "code_url": "https://huggingface.co/fastino/GLiDE/tree/main/glide", "base_model": "Qwen/Qwen3.8-27B", "base_url": "https://huggingface.co/Qwen/Qwen3.8-27B", "weights_repo": "fastino/GLiDE", "served_checkpoint": "a7e45896be7c96ba3354891da790c5ec99c30b6c", "weights_note": null, "org": {"id": "fastino", "name": "Fastino", "logo": "data/orgs/fastino.png"}, "pr": 255, "replaces": "decision-27b-ckpt-nothink", "new": true}. New in the 2026-10-21 PR batch (PR #255). Replaces the earlier board entry decision-27b-ckpt-nothink (newer version of the same model). Submitted public 0.3 run adopted (kit rescore 71.99). Private tests run by the lab on one RTX PRO 6000 with the author's serving; decision agreement with the submitted rows 99.90% on 895 rows (bar 98.5%). Weights are public on the Hugging Face Hub (fastino-community license); the lab scored revision a7e4589. Decision Index 0.4.1 (release-v4), source generated 2026-10-29T14:04:16Z. Full = 60% equated private same-skill + 40% equated private new-domain tasks; public contributes zero. Both private components are mapped onto the public scale using untuned stock models. Public weighting/panel changed (GSM8K rebuilt, ForecastBench retired, WinoGrande area reassigned); these are new benchmark definitions, not proof of a model upgrade. Public 37-benchmark index and 42 published native benchmarks remain separate. Preserve source rounding, coverage, exclusions and unknown execution dates; no new uncertainty inferred. Source notes: New in the 2026-10-21 PR batch (PR #255). Replaces the earlier board entry decision-27b-ckpt-nothink (newer version of the same model). Submitted public 0.3 run adopted (kit rescore 71.99). Private tests run by the lab on one RTX PRO 6000 with the author's serving; decision agreement with the submitted rows 99.90% on 895 rows (bar 98.5%). Weights are public on the Hugging Face Hub (fastino-community license); the lab scored revision a7e4589. Author public-run adoption/execution status and private lab setup are disclosed in configuration; evaluator-published does not mean every public run was independently executed. Published weighted contribution; not a distinct benchmark execution. The board served a7e45896be; the registry pins the newest repository revision, and the commit titles after it concern the card. Community Decision Index 0.4.1: expanded/refreshed harder private new-domain tasks and equating; Full remains 60% private same-skill +40% private new domains, zero public contribution. This population revision is not a model upgrade/regression. Public/native/calibration data remain on their unchanged 0.4 definitions. Source publication/generation is not a model run date. Public submitted runs plus evaluator private measurements are Evaluator-published, not wholly independently executed. Checked / reported: 2026-11-02 · Samples: Not published Snapshot: 2026-10-22T05:40:06Z · Revision: d9962113a1acc95c2d7522c2babfb989afbeb034; c130faf65f82e81d3bdefb01e758c712dc5c5199 for the 2 November 2026 additions · Published observations; result reuse terms not specified Result source · Benchmark source
HF text 0.4 · Knowledge · raw	68.16	Setup & source HF text 0.4 · Knowledge · raw Exact source engine glide-v1.2; {"kind": "full fine-tune", "technique": "autoregressive", "technique_name": "autoregressive", "served_params": 27800000000, "params_estimated": false, "params_from_name": false, "trained_bytes": null, "model_url": "https://huggingface.co/fastino/GLiDE", "code_url": "https://huggingface.co/fastino/GLiDE/tree/main/glide", "base_model": "Qwen/Qwen3.8-27B", "base_url": "https://huggingface.co/Qwen/Qwen3.8-27B", "weights_repo": "fastino/GLiDE", "served_checkpoint": "a7e45896be7c96ba3354891da790c5ec99c30b6c", "weights_note": null, "org": {"id": "fastino", "name": "Fastino", "logo": "data/orgs/fastino.png"}, "pr": 255, "replaces": "decision-27b-ckpt-nothink", "new": true}. New in the 2026-10-21 PR batch (PR #255). Replaces the earlier board entry decision-27b-ckpt-nothink (newer version of the same model). Submitted public 0.3 run adopted (kit rescore 71.99). Private tests run by the lab on one RTX PRO 6000 with the author's serving; decision agreement with the submitted rows 99.90% on 895 rows (bar 98.5%). Weights are public on the Hugging Face Hub (fastino-community license); the lab scored revision a7e4589. Decision Index 0.4.1 (release-v4), source generated 2026-10-29T14:04:16Z. Full = 60% equated private same-skill + 40% equated private new-domain tasks; public contributes zero. Both private components are mapped onto the public scale using untuned stock models. Public weighting/panel changed (GSM8K rebuilt, ForecastBench retired, WinoGrande area reassigned); these are new benchmark definitions, not proof of a model upgrade. Public 37-benchmark index and 42 published native benchmarks remain separate. Preserve source rounding, coverage, exclusions and unknown execution dates; no new uncertainty inferred. Source notes: New in the 2026-10-21 PR batch (PR #255). Replaces the earlier board entry decision-27b-ckpt-nothink (newer version of the same model). Submitted public 0.3 run adopted (kit rescore 71.99). Private tests run by the lab on one RTX PRO 6000 with the author's serving; decision agreement with the submitted rows 99.90% on 895 rows (bar 98.5%). Weights are public on the Hugging Face Hub (fastino-community license); the lab scored revision a7e4589. Author public-run adoption/execution status and private lab setup are disclosed in configuration; evaluator-published does not mean every public run was independently executed. Count is contributing benchmarks, not requests. Source area context: {"id": "knowledge", "raw": 0.6816, "skill": 0.6059, "combined": 0.5727, "coverage": 1, "pending": 0, "n": 11} The board served a7e45896be; the registry pins the newest repository revision, and the commit titles after it concern the card. Decision Index 0.4 (release-v4), source generated 2026-10-20T14:13:56Z. Full = 60% equated private same-skill + 40% equated private new-domain tasks; public contributes zero. Both private components are mapped onto the public scale using untuned stock models. Public weighting/panel changed (GSM8K rebuilt, ForecastBench retired, WinoGrande area reassigned); these are new benchmark definitions, not proof of a model upgrade. Public 37-benchmark index and 42 published native benchmarks remain separate. Preserve source rounding, coverage, exclusions and unknown execution dates; no new uncertainty inferred. Checked / reported: 2026-11-02 · Samples: 11 Snapshot: 2026-10-20T14:13:56Z · Revision: 133313cf8963f2b6bed960abb0152fd14cbb6731; c130faf65f82e81d3bdefb01e758c712dc5c5199 for the 2 November 2026 additions · Published observations; result reuse terms not specified Result source · Benchmark source
HF text 0.4 · Knowledge · skill	60.59	Setup & source HF text 0.4 · Knowledge · skill Exact source engine glide-v1.2; {"kind": "full fine-tune", "technique": "autoregressive", "technique_name": "autoregressive", "served_params": 27800000000, "params_estimated": false, "params_from_name": false, "trained_bytes": null, "model_url": "https://huggingface.co/fastino/GLiDE", "code_url": "https://huggingface.co/fastino/GLiDE/tree/main/glide", "base_model": "Qwen/Qwen3.8-27B", "base_url": "https://huggingface.co/Qwen/Qwen3.8-27B", "weights_repo": "fastino/GLiDE", "served_checkpoint": "a7e45896be7c96ba3354891da790c5ec99c30b6c", "weights_note": null, "org": {"id": "fastino", "name": "Fastino", "logo": "data/orgs/fastino.png"}, "pr": 255, "replaces": "decision-27b-ckpt-nothink", "new": true}. New in the 2026-10-21 PR batch (PR #255). Replaces the earlier board entry decision-27b-ckpt-nothink (newer version of the same model). Submitted public 0.3 run adopted (kit rescore 71.99). Private tests run by the lab on one RTX PRO 6000 with the author's serving; decision agreement with the submitted rows 99.90% on 895 rows (bar 98.5%). Weights are public on the Hugging Face Hub (fastino-community license); the lab scored revision a7e4589. Decision Index 0.4.1 (release-v4), source generated 2026-10-29T14:04:16Z. Full = 60% equated private same-skill + 40% equated private new-domain tasks; public contributes zero. Both private components are mapped onto the public scale using untuned stock models. Public weighting/panel changed (GSM8K rebuilt, ForecastBench retired, WinoGrande area reassigned); these are new benchmark definitions, not proof of a model upgrade. Public 37-benchmark index and 42 published native benchmarks remain separate. Preserve source rounding, coverage, exclusions and unknown execution dates; no new uncertainty inferred. Source notes: New in the 2026-10-21 PR batch (PR #255). Replaces the earlier board entry decision-27b-ckpt-nothink (newer version of the same model). Submitted public 0.3 run adopted (kit rescore 71.99). Private tests run by the lab on one RTX PRO 6000 with the author's serving; decision agreement with the submitted rows 99.90% on 895 rows (bar 98.5%). Weights are public on the Hugging Face Hub (fastino-community license); the lab scored revision a7e4589. Author public-run adoption/execution status and private lab setup are disclosed in configuration; evaluator-published does not mean every public run was independently executed. Count is contributing benchmarks, not requests. Source area context: {"id": "knowledge", "raw": 0.6816, "skill": 0.6059, "combined": 0.5727, "coverage": 1, "pending": 0, "n": 11} The board served a7e45896be; the registry pins the newest repository revision, and the commit titles after it concern the card. Decision Index 0.4 (release-v4), source generated 2026-10-20T14:13:56Z. Full = 60% equated private same-skill + 40% equated private new-domain tasks; public contributes zero. Both private components are mapped onto the public scale using untuned stock models. Public weighting/panel changed (GSM8K rebuilt, ForecastBench retired, WinoGrande area reassigned); these are new benchmark definitions, not proof of a model upgrade. Public 37-benchmark index and 42 published native benchmarks remain separate. Preserve source rounding, coverage, exclusions and unknown execution dates; no new uncertainty inferred. Checked / reported: 2026-11-02 · Samples: 11 Snapshot: 2026-10-20T14:13:56Z · Revision: 133313cf8963f2b6bed960abb0152fd14cbb6731; c130faf65f82e81d3bdefb01e758c712dc5c5199 for the 2 November 2026 additions · Published observations; result reuse terms not specified Result source · Benchmark source
HF text 0.4 · Knowledge · combined	57.27	Setup & source HF text 0.4 · Knowledge · combined Exact source engine glide-v1.2; {"kind": "full fine-tune", "technique": "autoregressive", "technique_name": "autoregressive", "served_params": 27800000000, "params_estimated": false, "params_from_name": false, "trained_bytes": null, "model_url": "https://huggingface.co/fastino/GLiDE", "code_url": "https://huggingface.co/fastino/GLiDE/tree/main/glide", "base_model": "Qwen/Qwen3.8-27B", "base_url": "https://huggingface.co/Qwen/Qwen3.8-27B", "weights_repo": "fastino/GLiDE", "served_checkpoint": "a7e45896be7c96ba3354891da790c5ec99c30b6c", "weights_note": null, "org": {"id": "fastino", "name": "Fastino", "logo": "data/orgs/fastino.png"}, "pr": 255, "replaces": "decision-27b-ckpt-nothink", "new": true}. New in the 2026-10-21 PR batch (PR #255). Replaces the earlier board entry decision-27b-ckpt-nothink (newer version of the same model). Submitted public 0.3 run adopted (kit rescore 71.99). Private tests run by the lab on one RTX PRO 6000 with the author's serving; decision agreement with the submitted rows 99.90% on 895 rows (bar 98.5%). Weights are public on the Hugging Face Hub (fastino-community license); the lab scored revision a7e4589. Decision Index 0.4.1 (release-v4), source generated 2026-10-29T14:04:16Z. Full = 60% equated private same-skill + 40% equated private new-domain tasks; public contributes zero. Both private components are mapped onto the public scale using untuned stock models. Public weighting/panel changed (GSM8K rebuilt, ForecastBench retired, WinoGrande area reassigned); these are new benchmark definitions, not proof of a model upgrade. Public 37-benchmark index and 42 published native benchmarks remain separate. Preserve source rounding, coverage, exclusions and unknown execution dates; no new uncertainty inferred. Source notes: New in the 2026-10-21 PR batch (PR #255). Replaces the earlier board entry decision-27b-ckpt-nothink (newer version of the same model). Submitted public 0.3 run adopted (kit rescore 71.99). Private tests run by the lab on one RTX PRO 6000 with the author's serving; decision agreement with the submitted rows 99.90% on 895 rows (bar 98.5%). Weights are public on the Hugging Face Hub (fastino-community license); the lab scored revision a7e4589. Author public-run adoption/execution status and private lab setup are disclosed in configuration; evaluator-published does not mean every public run was independently executed. Count is contributing benchmarks, not requests. Source area context: {"id": "knowledge", "raw": 0.6816, "skill": 0.6059, "combined": 0.5727, "coverage": 1, "pending": 0, "n": 11} The board served a7e45896be; the registry pins the newest repository revision, and the commit titles after it concern the card. Decision Index 0.4 (release-v4), source generated 2026-10-20T14:13:56Z. Full = 60% equated private same-skill + 40% equated private new-domain tasks; public contributes zero. Both private components are mapped onto the public scale using untuned stock models. Public weighting/panel changed (GSM8K rebuilt, ForecastBench retired, WinoGrande area reassigned); these are new benchmark definitions, not proof of a model upgrade. Public 37-benchmark index and 42 published native benchmarks remain separate. Preserve source rounding, coverage, exclusions and unknown execution dates; no new uncertainty inferred. Checked / reported: 2026-11-02 · Samples: 11 Snapshot: 2026-10-20T14:13:56Z · Revision: 133313cf8963f2b6bed960abb0152fd14cbb6731; c130faf65f82e81d3bdefb01e758c712dc5c5199 for the 2 November 2026 additions · Published observations; result reuse terms not specified Result source · Benchmark source
HF text 0.4 · Language · raw	82.85	Setup & source HF text 0.4 · Language · raw Exact source engine glide-v1.2; {"kind": "full fine-tune", "technique": "autoregressive", "technique_name": "autoregressive", "served_params": 27800000000, "params_estimated": false, "params_from_name": false, "trained_bytes": null, "model_url": "https://huggingface.co/fastino/GLiDE", "code_url": "https://huggingface.co/fastino/GLiDE/tree/main/glide", "base_model": "Qwen/Qwen3.8-27B", "base_url": "https://huggingface.co/Qwen/Qwen3.8-27B", "weights_repo": "fastino/GLiDE", "served_checkpoint": "a7e45896be7c96ba3354891da790c5ec99c30b6c", "weights_note": null, "org": {"id": "fastino", "name": "Fastino", "logo": "data/orgs/fastino.png"}, "pr": 255, "replaces": "decision-27b-ckpt-nothink", "new": true}. New in the 2026-10-21 PR batch (PR #255). Replaces the earlier board entry decision-27b-ckpt-nothink (newer version of the same model). Submitted public 0.3 run adopted (kit rescore 71.99). Private tests run by the lab on one RTX PRO 6000 with the author's serving; decision agreement with the submitted rows 99.90% on 895 rows (bar 98.5%). Weights are public on the Hugging Face Hub (fastino-community license); the lab scored revision a7e4589. Decision Index 0.4.1 (release-v4), source generated 2026-10-29T14:04:16Z. Full = 60% equated private same-skill + 40% equated private new-domain tasks; public contributes zero. Both private components are mapped onto the public scale using untuned stock models. Public weighting/panel changed (GSM8K rebuilt, ForecastBench retired, WinoGrande area reassigned); these are new benchmark definitions, not proof of a model upgrade. Public 37-benchmark index and 42 published native benchmarks remain separate. Preserve source rounding, coverage, exclusions and unknown execution dates; no new uncertainty inferred. Source notes: New in the 2026-10-21 PR batch (PR #255). Replaces the earlier board entry decision-27b-ckpt-nothink (newer version of the same model). Submitted public 0.3 run adopted (kit rescore 71.99). Private tests run by the lab on one RTX PRO 6000 with the author's serving; decision agreement with the submitted rows 99.90% on 895 rows (bar 98.5%). Weights are public on the Hugging Face Hub (fastino-community license); the lab scored revision a7e4589. Author public-run adoption/execution status and private lab setup are disclosed in configuration; evaluator-published does not mean every public run was independently executed. Count is contributing benchmarks, not requests. Source area context: {"id": "language", "raw": 0.8285, "skill": 0.7549, "combined": 0.6564, "coverage": 1, "pending": 0, "n": 9} The board served a7e45896be; the registry pins the newest repository revision, and the commit titles after it concern the card. Decision Index 0.4 (release-v4), source generated 2026-10-20T14:13:56Z. Full = 60% equated private same-skill + 40% equated private new-domain tasks; public contributes zero. Both private components are mapped onto the public scale using untuned stock models. Public weighting/panel changed (GSM8K rebuilt, ForecastBench retired, WinoGrande area reassigned); these are new benchmark definitions, not proof of a model upgrade. Public 37-benchmark index and 42 published native benchmarks remain separate. Preserve source rounding, coverage, exclusions and unknown execution dates; no new uncertainty inferred. Checked / reported: 2026-11-02 · Samples: 9 Snapshot: 2026-10-20T14:13:56Z · Revision: 133313cf8963f2b6bed960abb0152fd14cbb6731; c130faf65f82e81d3bdefb01e758c712dc5c5199 for the 2 November 2026 additions · Published observations; result reuse terms not specified Result source · Benchmark source
HF text 0.4 · Language · skill	75.49	Setup & source HF text 0.4 · Language · skill Exact source engine glide-v1.2; {"kind": "full fine-tune", "technique": "autoregressive", "technique_name": "autoregressive", "served_params": 27800000000, "params_estimated": false, "params_from_name": false, "trained_bytes": null, "model_url": "https://huggingface.co/fastino/GLiDE", "code_url": "https://huggingface.co/fastino/GLiDE/tree/main/glide", "base_model": "Qwen/Qwen3.8-27B", "base_url": "https://huggingface.co/Qwen/Qwen3.8-27B", "weights_repo": "fastino/GLiDE", "served_checkpoint": "a7e45896be7c96ba3354891da790c5ec99c30b6c", "weights_note": null, "org": {"id": "fastino", "name": "Fastino", "logo": "data/orgs/fastino.png"}, "pr": 255, "replaces": "decision-27b-ckpt-nothink", "new": true}. New in the 2026-10-21 PR batch (PR #255). Replaces the earlier board entry decision-27b-ckpt-nothink (newer version of the same model). Submitted public 0.3 run adopted (kit rescore 71.99). Private tests run by the lab on one RTX PRO 6000 with the author's serving; decision agreement with the submitted rows 99.90% on 895 rows (bar 98.5%). Weights are public on the Hugging Face Hub (fastino-community license); the lab scored revision a7e4589. Decision Index 0.4.1 (release-v4), source generated 2026-10-29T14:04:16Z. Full = 60% equated private same-skill + 40% equated private new-domain tasks; public contributes zero. Both private components are mapped onto the public scale using untuned stock models. Public weighting/panel changed (GSM8K rebuilt, ForecastBench retired, WinoGrande area reassigned); these are new benchmark definitions, not proof of a model upgrade. Public 37-benchmark index and 42 published native benchmarks remain separate. Preserve source rounding, coverage, exclusions and unknown execution dates; no new uncertainty inferred. Source notes: New in the 2026-10-21 PR batch (PR #255). Replaces the earlier board entry decision-27b-ckpt-nothink (newer version of the same model). Submitted public 0.3 run adopted (kit rescore 71.99). Private tests run by the lab on one RTX PRO 6000 with the author's serving; decision agreement with the submitted rows 99.90% on 895 rows (bar 98.5%). Weights are public on the Hugging Face Hub (fastino-community license); the lab scored revision a7e4589. Author public-run adoption/execution status and private lab setup are disclosed in configuration; evaluator-published does not mean every public run was independently executed. Count is contributing benchmarks, not requests. Source area context: {"id": "language", "raw": 0.8285, "skill": 0.7549, "combined": 0.6564, "coverage": 1, "pending": 0, "n": 9} The board served a7e45896be; the registry pins the newest repository revision, and the commit titles after it concern the card. Decision Index 0.4 (release-v4), source generated 2026-10-20T14:13:56Z. Full = 60% equated private same-skill + 40% equated private new-domain tasks; public contributes zero. Both private components are mapped onto the public scale using untuned stock models. Public weighting/panel changed (GSM8K rebuilt, ForecastBench retired, WinoGrande area reassigned); these are new benchmark definitions, not proof of a model upgrade. Public 37-benchmark index and 42 published native benchmarks remain separate. Preserve source rounding, coverage, exclusions and unknown execution dates; no new uncertainty inferred. Checked / reported: 2026-11-02 · Samples: 9 Snapshot: 2026-10-20T14:13:56Z · Revision: 133313cf8963f2b6bed960abb0152fd14cbb6731; c130faf65f82e81d3bdefb01e758c712dc5c5199 for the 2 November 2026 additions · Published observations; result reuse terms not specified Result source · Benchmark source
HF text 0.4 · Language · combined	65.64	Setup & source HF text 0.4 · Language · combined Exact source engine glide-v1.2; {"kind": "full fine-tune", "technique": "autoregressive", "technique_name": "autoregressive", "served_params": 27800000000, "params_estimated": false, "params_from_name": false, "trained_bytes": null, "model_url": "https://huggingface.co/fastino/GLiDE", "code_url": "https://huggingface.co/fastino/GLiDE/tree/main/glide", "base_model": "Qwen/Qwen3.8-27B", "base_url": "https://huggingface.co/Qwen/Qwen3.8-27B", "weights_repo": "fastino/GLiDE", "served_checkpoint": "a7e45896be7c96ba3354891da790c5ec99c30b6c", "weights_note": null, "org": {"id": "fastino", "name": "Fastino", "logo": "data/orgs/fastino.png"}, "pr": 255, "replaces": "decision-27b-ckpt-nothink", "new": true}. New in the 2026-10-21 PR batch (PR #255). Replaces the earlier board entry decision-27b-ckpt-nothink (newer version of the same model). Submitted public 0.3 run adopted (kit rescore 71.99). Private tests run by the lab on one RTX PRO 6000 with the author's serving; decision agreement with the submitted rows 99.90% on 895 rows (bar 98.5%). Weights are public on the Hugging Face Hub (fastino-community license); the lab scored revision a7e4589. Decision Index 0.4.1 (release-v4), source generated 2026-10-29T14:04:16Z. Full = 60% equated private same-skill + 40% equated private new-domain tasks; public contributes zero. Both private components are mapped onto the public scale using untuned stock models. Public weighting/panel changed (GSM8K rebuilt, ForecastBench retired, WinoGrande area reassigned); these are new benchmark definitions, not proof of a model upgrade. Public 37-benchmark index and 42 published native benchmarks remain separate. Preserve source rounding, coverage, exclusions and unknown execution dates; no new uncertainty inferred. Source notes: New in the 2026-10-21 PR batch (PR #255). Replaces the earlier board entry decision-27b-ckpt-nothink (newer version of the same model). Submitted public 0.3 run adopted (kit rescore 71.99). Private tests run by the lab on one RTX PRO 6000 with the author's serving; decision agreement with the submitted rows 99.90% on 895 rows (bar 98.5%). Weights are public on the Hugging Face Hub (fastino-community license); the lab scored revision a7e4589. Author public-run adoption/execution status and private lab setup are disclosed in configuration; evaluator-published does not mean every public run was independently executed. Count is contributing benchmarks, not requests. Source area context: {"id": "language", "raw": 0.8285, "skill": 0.7549, "combined": 0.6564, "coverage": 1, "pending": 0, "n": 9} The board served a7e45896be; the registry pins the newest repository revision, and the commit titles after it concern the card. Decision Index 0.4 (release-v4), source generated 2026-10-20T14:13:56Z. Full = 60% equated private same-skill + 40% equated private new-domain tasks; public contributes zero. Both private components are mapped onto the public scale using untuned stock models. Public weighting/panel changed (GSM8K rebuilt, ForecastBench retired, WinoGrande area reassigned); these are new benchmark definitions, not proof of a model upgrade. Public 37-benchmark index and 42 published native benchmarks remain separate. Preserve source rounding, coverage, exclusions and unknown execution dates; no new uncertainty inferred. Checked / reported: 2026-11-02 · Samples: 9 Snapshot: 2026-10-20T14:13:56Z · Revision: 133313cf8963f2b6bed960abb0152fd14cbb6731; c130faf65f82e81d3bdefb01e758c712dc5c5199 for the 2 November 2026 additions · Published observations; result reuse terms not specified Result source · Benchmark source
HF text 0.4 · Retrieval · raw	78.79	Setup & source HF text 0.4 · Retrieval · raw Exact source engine glide-v1.2; {"kind": "full fine-tune", "technique": "autoregressive", "technique_name": "autoregressive", "served_params": 27800000000, "params_estimated": false, "params_from_name": false, "trained_bytes": null, "model_url": "https://huggingface.co/fastino/GLiDE", "code_url": "https://huggingface.co/fastino/GLiDE/tree/main/glide", "base_model": "Qwen/Qwen3.8-27B", "base_url": "https://huggingface.co/Qwen/Qwen3.8-27B", "weights_repo": "fastino/GLiDE", "served_checkpoint": "a7e45896be7c96ba3354891da790c5ec99c30b6c", "weights_note": null, "org": {"id": "fastino", "name": "Fastino", "logo": "data/orgs/fastino.png"}, "pr": 255, "replaces": "decision-27b-ckpt-nothink", "new": true}. New in the 2026-10-21 PR batch (PR #255). Replaces the earlier board entry decision-27b-ckpt-nothink (newer version of the same model). Submitted public 0.3 run adopted (kit rescore 71.99). Private tests run by the lab on one RTX PRO 6000 with the author's serving; decision agreement with the submitted rows 99.90% on 895 rows (bar 98.5%). Weights are public on the Hugging Face Hub (fastino-community license); the lab scored revision a7e4589. Decision Index 0.4.1 (release-v4), source generated 2026-10-29T14:04:16Z. Full = 60% equated private same-skill + 40% equated private new-domain tasks; public contributes zero. Both private components are mapped onto the public scale using untuned stock models. Public weighting/panel changed (GSM8K rebuilt, ForecastBench retired, WinoGrande area reassigned); these are new benchmark definitions, not proof of a model upgrade. Public 37-benchmark index and 42 published native benchmarks remain separate. Preserve source rounding, coverage, exclusions and unknown execution dates; no new uncertainty inferred. Source notes: New in the 2026-10-21 PR batch (PR #255). Replaces the earlier board entry decision-27b-ckpt-nothink (newer version of the same model). Submitted public 0.3 run adopted (kit rescore 71.99). Private tests run by the lab on one RTX PRO 6000 with the author's serving; decision agreement with the submitted rows 99.90% on 895 rows (bar 98.5%). Weights are public on the Hugging Face Hub (fastino-community license); the lab scored revision a7e4589. Author public-run adoption/execution status and private lab setup are disclosed in configuration; evaluator-published does not mean every public run was independently executed. Count is contributing benchmarks, not requests. Source area context: {"id": "retrieval", "raw": 0.7879, "skill": 0.7224, "combined": 0.7307, "coverage": 1, "pending": 0, "n": 6} The board served a7e45896be; the registry pins the newest repository revision, and the commit titles after it concern the card. Decision Index 0.4 (release-v4), source generated 2026-10-20T14:13:56Z. Full = 60% equated private same-skill + 40% equated private new-domain tasks; public contributes zero. Both private components are mapped onto the public scale using untuned stock models. Public weighting/panel changed (GSM8K rebuilt, ForecastBench retired, WinoGrande area reassigned); these are new benchmark definitions, not proof of a model upgrade. Public 37-benchmark index and 42 published native benchmarks remain separate. Preserve source rounding, coverage, exclusions and unknown execution dates; no new uncertainty inferred. Checked / reported: 2026-11-02 · Samples: 6 Snapshot: 2026-10-20T14:13:56Z · Revision: 133313cf8963f2b6bed960abb0152fd14cbb6731; c130faf65f82e81d3bdefb01e758c712dc5c5199 for the 2 November 2026 additions · Published observations; result reuse terms not specified Result source · Benchmark source
HF text 0.4 · Retrieval · skill	72.24	Setup & source HF text 0.4 · Retrieval · skill Exact source engine glide-v1.2; {"kind": "full fine-tune", "technique": "autoregressive", "technique_name": "autoregressive", "served_params": 27800000000, "params_estimated": false, "params_from_name": false, "trained_bytes": null, "model_url": "https://huggingface.co/fastino/GLiDE", "code_url": "https://huggingface.co/fastino/GLiDE/tree/main/glide", "base_model": "Qwen/Qwen3.8-27B", "base_url": "https://huggingface.co/Qwen/Qwen3.8-27B", "weights_repo": "fastino/GLiDE", "served_checkpoint": "a7e45896be7c96ba3354891da790c5ec99c30b6c", "weights_note": null, "org": {"id": "fastino", "name": "Fastino", "logo": "data/orgs/fastino.png"}, "pr": 255, "replaces": "decision-27b-ckpt-nothink", "new": true}. New in the 2026-10-21 PR batch (PR #255). Replaces the earlier board entry decision-27b-ckpt-nothink (newer version of the same model). Submitted public 0.3 run adopted (kit rescore 71.99). Private tests run by the lab on one RTX PRO 6000 with the author's serving; decision agreement with the submitted rows 99.90% on 895 rows (bar 98.5%). Weights are public on the Hugging Face Hub (fastino-community license); the lab scored revision a7e4589. Decision Index 0.4.1 (release-v4), source generated 2026-10-29T14:04:16Z. Full = 60% equated private same-skill + 40% equated private new-domain tasks; public contributes zero. Both private components are mapped onto the public scale using untuned stock models. Public weighting/panel changed (GSM8K rebuilt, ForecastBench retired, WinoGrande area reassigned); these are new benchmark definitions, not proof of a model upgrade. Public 37-benchmark index and 42 published native benchmarks remain separate. Preserve source rounding, coverage, exclusions and unknown execution dates; no new uncertainty inferred. Source notes: New in the 2026-10-21 PR batch (PR #255). Replaces the earlier board entry decision-27b-ckpt-nothink (newer version of the same model). Submitted public 0.3 run adopted (kit rescore 71.99). Private tests run by the lab on one RTX PRO 6000 with the author's serving; decision agreement with the submitted rows 99.90% on 895 rows (bar 98.5%). Weights are public on the Hugging Face Hub (fastino-community license); the lab scored revision a7e4589. Author public-run adoption/execution status and private lab setup are disclosed in configuration; evaluator-published does not mean every public run was independently executed. Count is contributing benchmarks, not requests. Source area context: {"id": "retrieval", "raw": 0.7879, "skill": 0.7224, "combined": 0.7307, "coverage": 1, "pending": 0, "n": 6} The board served a7e45896be; the registry pins the newest repository revision, and the commit titles after it concern the card. Decision Index 0.4 (release-v4), source generated 2026-10-20T14:13:56Z. Full = 60% equated private same-skill + 40% equated private new-domain tasks; public contributes zero. Both private components are mapped onto the public scale using untuned stock models. Public weighting/panel changed (GSM8K rebuilt, ForecastBench retired, WinoGrande area reassigned); these are new benchmark definitions, not proof of a model upgrade. Public 37-benchmark index and 42 published native benchmarks remain separate. Preserve source rounding, coverage, exclusions and unknown execution dates; no new uncertainty inferred. Checked / reported: 2026-11-02 · Samples: 6 Snapshot: 2026-10-20T14:13:56Z · Revision: 133313cf8963f2b6bed960abb0152fd14cbb6731; c130faf65f82e81d3bdefb01e758c712dc5c5199 for the 2 November 2026 additions · Published observations; result reuse terms not specified Result source · Benchmark source
HF text 0.4 · Retrieval · combined	73.07	Setup & source HF text 0.4 · Retrieval · combined Exact source engine glide-v1.2; {"kind": "full fine-tune", "technique": "autoregressive", "technique_name": "autoregressive", "served_params": 27800000000, "params_estimated": false, "params_from_name": false, "trained_bytes": null, "model_url": "https://huggingface.co/fastino/GLiDE", "code_url": "https://huggingface.co/fastino/GLiDE/tree/main/glide", "base_model": "Qwen/Qwen3.8-27B", "base_url": "https://huggingface.co/Qwen/Qwen3.8-27B", "weights_repo": "fastino/GLiDE", "served_checkpoint": "a7e45896be7c96ba3354891da790c5ec99c30b6c", "weights_note": null, "org": {"id": "fastino", "name": "Fastino", "logo": "data/orgs/fastino.png"}, "pr": 255, "replaces": "decision-27b-ckpt-nothink", "new": true}. New in the 2026-10-21 PR batch (PR #255). Replaces the earlier board entry decision-27b-ckpt-nothink (newer version of the same model). Submitted public 0.3 run adopted (kit rescore 71.99). Private tests run by the lab on one RTX PRO 6000 with the author's serving; decision agreement with the submitted rows 99.90% on 895 rows (bar 98.5%). Weights are public on the Hugging Face Hub (fastino-community license); the lab scored revision a7e4589. Decision Index 0.4.1 (release-v4), source generated 2026-10-29T14:04:16Z. Full = 60% equated private same-skill + 40% equated private new-domain tasks; public contributes zero. Both private components are mapped onto the public scale using untuned stock models. Public weighting/panel changed (GSM8K rebuilt, ForecastBench retired, WinoGrande area reassigned); these are new benchmark definitions, not proof of a model upgrade. Public 37-benchmark index and 42 published native benchmarks remain separate. Preserve source rounding, coverage, exclusions and unknown execution dates; no new uncertainty inferred. Source notes: New in the 2026-10-21 PR batch (PR #255). Replaces the earlier board entry decision-27b-ckpt-nothink (newer version of the same model). Submitted public 0.3 run adopted (kit rescore 71.99). Private tests run by the lab on one RTX PRO 6000 with the author's serving; decision agreement with the submitted rows 99.90% on 895 rows (bar 98.5%). Weights are public on the Hugging Face Hub (fastino-community license); the lab scored revision a7e4589. Author public-run adoption/execution status and private lab setup are disclosed in configuration; evaluator-published does not mean every public run was independently executed. Count is contributing benchmarks, not requests. Source area context: {"id": "retrieval", "raw": 0.7879, "skill": 0.7224, "combined": 0.7307, "coverage": 1, "pending": 0, "n": 6} The board served a7e45896be; the registry pins the newest repository revision, and the commit titles after it concern the card. Decision Index 0.4 (release-v4), source generated 2026-10-20T14:13:56Z. Full = 60% equated private same-skill + 40% equated private new-domain tasks; public contributes zero. Both private components are mapped onto the public scale using untuned stock models. Public weighting/panel changed (GSM8K rebuilt, ForecastBench retired, WinoGrande area reassigned); these are new benchmark definitions, not proof of a model upgrade. Public 37-benchmark index and 42 published native benchmarks remain separate. Preserve source rounding, coverage, exclusions and unknown execution dates; no new uncertainty inferred. Checked / reported: 2026-11-02 · Samples: 6 Snapshot: 2026-10-20T14:13:56Z · Revision: 133313cf8963f2b6bed960abb0152fd14cbb6731; c130faf65f82e81d3bdefb01e758c712dc5c5199 for the 2 November 2026 additions · Published observations; result reuse terms not specified Result source · Benchmark source
HF text 0.4 · Tools · raw	89.64	Setup & source HF text 0.4 · Tools · raw Exact source engine glide-v1.2; {"kind": "full fine-tune", "technique": "autoregressive", "technique_name": "autoregressive", "served_params": 27800000000, "params_estimated": false, "params_from_name": false, "trained_bytes": null, "model_url": "https://huggingface.co/fastino/GLiDE", "code_url": "https://huggingface.co/fastino/GLiDE/tree/main/glide", "base_model": "Qwen/Qwen3.8-27B", "base_url": "https://huggingface.co/Qwen/Qwen3.8-27B", "weights_repo": "fastino/GLiDE", "served_checkpoint": "a7e45896be7c96ba3354891da790c5ec99c30b6c", "weights_note": null, "org": {"id": "fastino", "name": "Fastino", "logo": "data/orgs/fastino.png"}, "pr": 255, "replaces": "decision-27b-ckpt-nothink", "new": true}. New in the 2026-10-21 PR batch (PR #255). Replaces the earlier board entry decision-27b-ckpt-nothink (newer version of the same model). Submitted public 0.3 run adopted (kit rescore 71.99). Private tests run by the lab on one RTX PRO 6000 with the author's serving; decision agreement with the submitted rows 99.90% on 895 rows (bar 98.5%). Weights are public on the Hugging Face Hub (fastino-community license); the lab scored revision a7e4589. Decision Index 0.4.1 (release-v4), source generated 2026-10-29T14:04:16Z. Full = 60% equated private same-skill + 40% equated private new-domain tasks; public contributes zero. Both private components are mapped onto the public scale using untuned stock models. Public weighting/panel changed (GSM8K rebuilt, ForecastBench retired, WinoGrande area reassigned); these are new benchmark definitions, not proof of a model upgrade. Public 37-benchmark index and 42 published native benchmarks remain separate. Preserve source rounding, coverage, exclusions and unknown execution dates; no new uncertainty inferred. Source notes: New in the 2026-10-21 PR batch (PR #255). Replaces the earlier board entry decision-27b-ckpt-nothink (newer version of the same model). Submitted public 0.3 run adopted (kit rescore 71.99). Private tests run by the lab on one RTX PRO 6000 with the author's serving; decision agreement with the submitted rows 99.90% on 895 rows (bar 98.5%). Weights are public on the Hugging Face Hub (fastino-community license); the lab scored revision a7e4589. Author public-run adoption/execution status and private lab setup are disclosed in configuration; evaluator-published does not mean every public run was independently executed. Count is contributing benchmarks, not requests. Source area context: {"id": "tools", "raw": 0.8964, "skill": 0.8804, "combined": 0.8491, "coverage": 1, "pending": 0, "n": 5} The board served a7e45896be; the registry pins the newest repository revision, and the commit titles after it concern the card. Decision Index 0.4 (release-v4), source generated 2026-10-20T14:13:56Z. Full = 60% equated private same-skill + 40% equated private new-domain tasks; public contributes zero. Both private components are mapped onto the public scale using untuned stock models. Public weighting/panel changed (GSM8K rebuilt, ForecastBench retired, WinoGrande area reassigned); these are new benchmark definitions, not proof of a model upgrade. Public 37-benchmark index and 42 published native benchmarks remain separate. Preserve source rounding, coverage, exclusions and unknown execution dates; no new uncertainty inferred. Checked / reported: 2026-11-02 · Samples: 5 Snapshot: 2026-10-20T14:13:56Z · Revision: 133313cf8963f2b6bed960abb0152fd14cbb6731; c130faf65f82e81d3bdefb01e758c712dc5c5199 for the 2 November 2026 additions · Published observations; result reuse terms not specified Result source · Benchmark source
HF text 0.4 · Tools · skill	88.04	Setup & source HF text 0.4 · Tools · skill Exact source engine glide-v1.2; {"kind": "full fine-tune", "technique": "autoregressive", "technique_name": "autoregressive", "served_params": 27800000000, "params_estimated": false, "params_from_name": false, "trained_bytes": null, "model_url": "https://huggingface.co/fastino/GLiDE", "code_url": "https://huggingface.co/fastino/GLiDE/tree/main/glide", "base_model": "Qwen/Qwen3.8-27B", "base_url": "https://huggingface.co/Qwen/Qwen3.8-27B", "weights_repo": "fastino/GLiDE", "served_checkpoint": "a7e45896be7c96ba3354891da790c5ec99c30b6c", "weights_note": null, "org": {"id": "fastino", "name": "Fastino", "logo": "data/orgs/fastino.png"}, "pr": 255, "replaces": "decision-27b-ckpt-nothink", "new": true}. New in the 2026-10-21 PR batch (PR #255). Replaces the earlier board entry decision-27b-ckpt-nothink (newer version of the same model). Submitted public 0.3 run adopted (kit rescore 71.99). Private tests run by the lab on one RTX PRO 6000 with the author's serving; decision agreement with the submitted rows 99.90% on 895 rows (bar 98.5%). Weights are public on the Hugging Face Hub (fastino-community license); the lab scored revision a7e4589. Decision Index 0.4.1 (release-v4), source generated 2026-10-29T14:04:16Z. Full = 60% equated private same-skill + 40% equated private new-domain tasks; public contributes zero. Both private components are mapped onto the public scale using untuned stock models. Public weighting/panel changed (GSM8K rebuilt, ForecastBench retired, WinoGrande area reassigned); these are new benchmark definitions, not proof of a model upgrade. Public 37-benchmark index and 42 published native benchmarks remain separate. Preserve source rounding, coverage, exclusions and unknown execution dates; no new uncertainty inferred. Source notes: New in the 2026-10-21 PR batch (PR #255). Replaces the earlier board entry decision-27b-ckpt-nothink (newer version of the same model). Submitted public 0.3 run adopted (kit rescore 71.99). Private tests run by the lab on one RTX PRO 6000 with the author's serving; decision agreement with the submitted rows 99.90% on 895 rows (bar 98.5%). Weights are public on the Hugging Face Hub (fastino-community license); the lab scored revision a7e4589. Author public-run adoption/execution status and private lab setup are disclosed in configuration; evaluator-published does not mean every public run was independently executed. Count is contributing benchmarks, not requests. Source area context: {"id": "tools", "raw": 0.8964, "skill": 0.8804, "combined": 0.8491, "coverage": 1, "pending": 0, "n": 5} The board served a7e45896be; the registry pins the newest repository revision, and the commit titles after it concern the card. Decision Index 0.4 (release-v4), source generated 2026-10-20T14:13:56Z. Full = 60% equated private same-skill + 40% equated private new-domain tasks; public contributes zero. Both private components are mapped onto the public scale using untuned stock models. Public weighting/panel changed (GSM8K rebuilt, ForecastBench retired, WinoGrande area reassigned); these are new benchmark definitions, not proof of a model upgrade. Public 37-benchmark index and 42 published native benchmarks remain separate. Preserve source rounding, coverage, exclusions and unknown execution dates; no new uncertainty inferred. Checked / reported: 2026-11-02 · Samples: 5 Snapshot: 2026-10-20T14:13:56Z · Revision: 133313cf8963f2b6bed960abb0152fd14cbb6731; c130faf65f82e81d3bdefb01e758c712dc5c5199 for the 2 November 2026 additions · Published observations; result reuse terms not specified Result source · Benchmark source
HF text 0.4 · Tools · combined	84.91	Setup & source HF text 0.4 · Tools · combined Exact source engine glide-v1.2; {"kind": "full fine-tune", "technique": "autoregressive", "technique_name": "autoregressive", "served_params": 27800000000, "params_estimated": false, "params_from_name": false, "trained_bytes": null, "model_url": "https://huggingface.co/fastino/GLiDE", "code_url": "https://huggingface.co/fastino/GLiDE/tree/main/glide", "base_model": "Qwen/Qwen3.8-27B", "base_url": "https://huggingface.co/Qwen/Qwen3.8-27B", "weights_repo": "fastino/GLiDE", "served_checkpoint": "a7e45896be7c96ba3354891da790c5ec99c30b6c", "weights_note": null, "org": {"id": "fastino", "name": "Fastino", "logo": "data/orgs/fastino.png"}, "pr": 255, "replaces": "decision-27b-ckpt-nothink", "new": true}. New in the 2026-10-21 PR batch (PR #255). Replaces the earlier board entry decision-27b-ckpt-nothink (newer version of the same model). Submitted public 0.3 run adopted (kit rescore 71.99). Private tests run by the lab on one RTX PRO 6000 with the author's serving; decision agreement with the submitted rows 99.90% on 895 rows (bar 98.5%). Weights are public on the Hugging Face Hub (fastino-community license); the lab scored revision a7e4589. Decision Index 0.4.1 (release-v4), source generated 2026-10-29T14:04:16Z. Full = 60% equated private same-skill + 40% equated private new-domain tasks; public contributes zero. Both private components are mapped onto the public scale using untuned stock models. Public weighting/panel changed (GSM8K rebuilt, ForecastBench retired, WinoGrande area reassigned); these are new benchmark definitions, not proof of a model upgrade. Public 37-benchmark index and 42 published native benchmarks remain separate. Preserve source rounding, coverage, exclusions and unknown execution dates; no new uncertainty inferred. Source notes: New in the 2026-10-21 PR batch (PR #255). Replaces the earlier board entry decision-27b-ckpt-nothink (newer version of the same model). Submitted public 0.3 run adopted (kit rescore 71.99). Private tests run by the lab on one RTX PRO 6000 with the author's serving; decision agreement with the submitted rows 99.90% on 895 rows (bar 98.5%). Weights are public on the Hugging Face Hub (fastino-community license); the lab scored revision a7e4589. Author public-run adoption/execution status and private lab setup are disclosed in configuration; evaluator-published does not mean every public run was independently executed. Count is contributing benchmarks, not requests. Source area context: {"id": "tools", "raw": 0.8964, "skill": 0.8804, "combined": 0.8491, "coverage": 1, "pending": 0, "n": 5} The board served a7e45896be; the registry pins the newest repository revision, and the commit titles after it concern the card. Decision Index 0.4 (release-v4), source generated 2026-10-20T14:13:56Z. Full = 60% equated private same-skill + 40% equated private new-domain tasks; public contributes zero. Both private components are mapped onto the public scale using untuned stock models. Public weighting/panel changed (GSM8K rebuilt, ForecastBench retired, WinoGrande area reassigned); these are new benchmark definitions, not proof of a model upgrade. Public 37-benchmark index and 42 published native benchmarks remain separate. Preserve source rounding, coverage, exclusions and unknown execution dates; no new uncertainty inferred. Checked / reported: 2026-11-02 · Samples: 5 Snapshot: 2026-10-20T14:13:56Z · Revision: 133313cf8963f2b6bed960abb0152fd14cbb6731; c130faf65f82e81d3bdefb01e758c712dc5c5199 for the 2 November 2026 additions · Published observations; result reuse terms not specified Result source · Benchmark source
HF text 0.4 · Arts · raw	76.90	Setup & source HF text 0.4 · Arts · raw Exact source engine glide-v1.2; {"kind": "full fine-tune", "technique": "autoregressive", "technique_name": "autoregressive", "served_params": 27800000000, "params_estimated": false, "params_from_name": false, "trained_bytes": null, "model_url": "https://huggingface.co/fastino/GLiDE", "code_url": "https://huggingface.co/fastino/GLiDE/tree/main/glide", "base_model": "Qwen/Qwen3.8-27B", "base_url": "https://huggingface.co/Qwen/Qwen3.8-27B", "weights_repo": "fastino/GLiDE", "served_checkpoint": "a7e45896be7c96ba3354891da790c5ec99c30b6c", "weights_note": null, "org": {"id": "fastino", "name": "Fastino", "logo": "data/orgs/fastino.png"}, "pr": 255, "replaces": "decision-27b-ckpt-nothink", "new": true}. New in the 2026-10-21 PR batch (PR #255). Replaces the earlier board entry decision-27b-ckpt-nothink (newer version of the same model). Submitted public 0.3 run adopted (kit rescore 71.99). Private tests run by the lab on one RTX PRO 6000 with the author's serving; decision agreement with the submitted rows 99.90% on 895 rows (bar 98.5%). Weights are public on the Hugging Face Hub (fastino-community license); the lab scored revision a7e4589. Decision Index 0.4.1 (release-v4), source generated 2026-10-29T14:04:16Z. Full = 60% equated private same-skill + 40% equated private new-domain tasks; public contributes zero. Both private components are mapped onto the public scale using untuned stock models. Public weighting/panel changed (GSM8K rebuilt, ForecastBench retired, WinoGrande area reassigned); these are new benchmark definitions, not proof of a model upgrade. Public 37-benchmark index and 42 published native benchmarks remain separate. Preserve source rounding, coverage, exclusions and unknown execution dates; no new uncertainty inferred. Source notes: New in the 2026-10-21 PR batch (PR #255). Replaces the earlier board entry decision-27b-ckpt-nothink (newer version of the same model). Submitted public 0.3 run adopted (kit rescore 71.99). Private tests run by the lab on one RTX PRO 6000 with the author's serving; decision agreement with the submitted rows 99.90% on 895 rows (bar 98.5%). Weights are public on the Hugging Face Hub (fastino-community license); the lab scored revision a7e4589. Author public-run adoption/execution status and private lab setup are disclosed in configuration; evaluator-published does not mean every public run was independently executed. Count is contributing benchmarks, not requests. Source area context: {"id": "arts", "raw": 0.769, "skill": 0.6259, "combined": 0.5867, "coverage": 1, "pending": 0, "n": 6} The board served a7e45896be; the registry pins the newest repository revision, and the commit titles after it concern the card. Decision Index 0.4 (release-v4), source generated 2026-10-20T14:13:56Z. Full = 60% equated private same-skill + 40% equated private new-domain tasks; public contributes zero. Both private components are mapped onto the public scale using untuned stock models. Public weighting/panel changed (GSM8K rebuilt, ForecastBench retired, WinoGrande area reassigned); these are new benchmark definitions, not proof of a model upgrade. Public 37-benchmark index and 42 published native benchmarks remain separate. Preserve source rounding, coverage, exclusions and unknown execution dates; no new uncertainty inferred. Checked / reported: 2026-11-02 · Samples: 6 Snapshot: 2026-10-20T14:13:56Z · Revision: 133313cf8963f2b6bed960abb0152fd14cbb6731; c130faf65f82e81d3bdefb01e758c712dc5c5199 for the 2 November 2026 additions · Published observations; result reuse terms not specified Result source · Benchmark source
HF text 0.4 · Arts · skill	62.59	Setup & source HF text 0.4 · Arts · skill Exact source engine glide-v1.2; {"kind": "full fine-tune", "technique": "autoregressive", "technique_name": "autoregressive", "served_params": 27800000000, "params_estimated": false, "params_from_name": false, "trained_bytes": null, "model_url": "https://huggingface.co/fastino/GLiDE", "code_url": "https://huggingface.co/fastino/GLiDE/tree/main/glide", "base_model": "Qwen/Qwen3.8-27B", "base_url": "https://huggingface.co/Qwen/Qwen3.8-27B", "weights_repo": "fastino/GLiDE", "served_checkpoint": "a7e45896be7c96ba3354891da790c5ec99c30b6c", "weights_note": null, "org": {"id": "fastino", "name": "Fastino", "logo": "data/orgs/fastino.png"}, "pr": 255, "replaces": "decision-27b-ckpt-nothink", "new": true}. New in the 2026-10-21 PR batch (PR #255). Replaces the earlier board entry decision-27b-ckpt-nothink (newer version of the same model). Submitted public 0.3 run adopted (kit rescore 71.99). Private tests run by the lab on one RTX PRO 6000 with the author's serving; decision agreement with the submitted rows 99.90% on 895 rows (bar 98.5%). Weights are public on the Hugging Face Hub (fastino-community license); the lab scored revision a7e4589. Decision Index 0.4.1 (release-v4), source generated 2026-10-29T14:04:16Z. Full = 60% equated private same-skill + 40% equated private new-domain tasks; public contributes zero. Both private components are mapped onto the public scale using untuned stock models. Public weighting/panel changed (GSM8K rebuilt, ForecastBench retired, WinoGrande area reassigned); these are new benchmark definitions, not proof of a model upgrade. Public 37-benchmark index and 42 published native benchmarks remain separate. Preserve source rounding, coverage, exclusions and unknown execution dates; no new uncertainty inferred. Source notes: New in the 2026-10-21 PR batch (PR #255). Replaces the earlier board entry decision-27b-ckpt-nothink (newer version of the same model). Submitted public 0.3 run adopted (kit rescore 71.99). Private tests run by the lab on one RTX PRO 6000 with the author's serving; decision agreement with the submitted rows 99.90% on 895 rows (bar 98.5%). Weights are public on the Hugging Face Hub (fastino-community license); the lab scored revision a7e4589. Author public-run adoption/execution status and private lab setup are disclosed in configuration; evaluator-published does not mean every public run was independently executed. Count is contributing benchmarks, not requests. Source area context: {"id": "arts", "raw": 0.769, "skill": 0.6259, "combined": 0.5867, "coverage": 1, "pending": 0, "n": 6} The board served a7e45896be; the registry pins the newest repository revision, and the commit titles after it concern the card. Decision Index 0.4 (release-v4), source generated 2026-10-20T14:13:56Z. Full = 60% equated private same-skill + 40% equated private new-domain tasks; public contributes zero. Both private components are mapped onto the public scale using untuned stock models. Public weighting/panel changed (GSM8K rebuilt, ForecastBench retired, WinoGrande area reassigned); these are new benchmark definitions, not proof of a model upgrade. Public 37-benchmark index and 42 published native benchmarks remain separate. Preserve source rounding, coverage, exclusions and unknown execution dates; no new uncertainty inferred. Checked / reported: 2026-11-02 · Samples: 6 Snapshot: 2026-10-20T14:13:56Z · Revision: 133313cf8963f2b6bed960abb0152fd14cbb6731; c130faf65f82e81d3bdefb01e758c712dc5c5199 for the 2 November 2026 additions · Published observations; result reuse terms not specified Result source · Benchmark source
HF text 0.4 · Arts · combined	58.67	Setup & source HF text 0.4 · Arts · combined Exact source engine glide-v1.2; {"kind": "full fine-tune", "technique": "autoregressive", "technique_name": "autoregressive", "served_params": 27800000000, "params_estimated": false, "params_from_name": false, "trained_bytes": null, "model_url": "https://huggingface.co/fastino/GLiDE", "code_url": "https://huggingface.co/fastino/GLiDE/tree/main/glide", "base_model": "Qwen/Qwen3.8-27B", "base_url": "https://huggingface.co/Qwen/Qwen3.8-27B", "weights_repo": "fastino/GLiDE", "served_checkpoint": "a7e45896be7c96ba3354891da790c5ec99c30b6c", "weights_note": null, "org": {"id": "fastino", "name": "Fastino", "logo": "data/orgs/fastino.png"}, "pr": 255, "replaces": "decision-27b-ckpt-nothink", "new": true}. New in the 2026-10-21 PR batch (PR #255). Replaces the earlier board entry decision-27b-ckpt-nothink (newer version of the same model). Submitted public 0.3 run adopted (kit rescore 71.99). Private tests run by the lab on one RTX PRO 6000 with the author's serving; decision agreement with the submitted rows 99.90% on 895 rows (bar 98.5%). Weights are public on the Hugging Face Hub (fastino-community license); the lab scored revision a7e4589. Decision Index 0.4.1 (release-v4), source generated 2026-10-29T14:04:16Z. Full = 60% equated private same-skill + 40% equated private new-domain tasks; public contributes zero. Both private components are mapped onto the public scale using untuned stock models. Public weighting/panel changed (GSM8K rebuilt, ForecastBench retired, WinoGrande area reassigned); these are new benchmark definitions, not proof of a model upgrade. Public 37-benchmark index and 42 published native benchmarks remain separate. Preserve source rounding, coverage, exclusions and unknown execution dates; no new uncertainty inferred. Source notes: New in the 2026-10-21 PR batch (PR #255). Replaces the earlier board entry decision-27b-ckpt-nothink (newer version of the same model). Submitted public 0.3 run adopted (kit rescore 71.99). Private tests run by the lab on one RTX PRO 6000 with the author's serving; decision agreement with the submitted rows 99.90% on 895 rows (bar 98.5%). Weights are public on the Hugging Face Hub (fastino-community license); the lab scored revision a7e4589. Author public-run adoption/execution status and private lab setup are disclosed in configuration; evaluator-published does not mean every public run was independently executed. Count is contributing benchmarks, not requests. Source area context: {"id": "arts", "raw": 0.769, "skill": 0.6259, "combined": 0.5867, "coverage": 1, "pending": 0, "n": 6} The board served a7e45896be; the registry pins the newest repository revision, and the commit titles after it concern the card. Decision Index 0.4 (release-v4), source generated 2026-10-20T14:13:56Z. Full = 60% equated private same-skill + 40% equated private new-domain tasks; public contributes zero. Both private components are mapped onto the public scale using untuned stock models. Public weighting/panel changed (GSM8K rebuilt, ForecastBench retired, WinoGrande area reassigned); these are new benchmark definitions, not proof of a model upgrade. Public 37-benchmark index and 42 published native benchmarks remain separate. Preserve source rounding, coverage, exclusions and unknown execution dates; no new uncertainty inferred. Checked / reported: 2026-11-02 · Samples: 6 Snapshot: 2026-10-20T14:13:56Z · Revision: 133313cf8963f2b6bed960abb0152fd14cbb6731; c130faf65f82e81d3bdefb01e758c712dc5c5199 for the 2 November 2026 additions · Published observations; result reuse terms not specified Result source · Benchmark source
HF text calibration 0.4 · ECE	2.04 pp	Setup & source HF text calibration 0.4 · ECE Exact source engine glide-v1.2; {"kind": "full fine-tune", "technique": "autoregressive", "technique_name": "autoregressive", "served_params": 27800000000, "params_estimated": false, "params_from_name": false, "trained_bytes": null, "model_url": "https://huggingface.co/fastino/GLiDE", "code_url": "https://huggingface.co/fastino/GLiDE/tree/main/glide", "base_model": "Qwen/Qwen3.8-27B", "base_url": "https://huggingface.co/Qwen/Qwen3.8-27B", "weights_repo": "fastino/GLiDE", "served_checkpoint": "a7e45896be7c96ba3354891da790c5ec99c30b6c", "weights_note": null, "org": {"id": "fastino", "name": "Fastino", "logo": "data/orgs/fastino.png"}, "pr": 255, "replaces": "decision-27b-ckpt-nothink", "new": true}. New in the 2026-10-21 PR batch (PR #255). Replaces the earlier board entry decision-27b-ckpt-nothink (newer version of the same model). Submitted public 0.3 run adopted (kit rescore 71.99). Private tests run by the lab on one RTX PRO 6000 with the author's serving; decision agreement with the submitted rows 99.90% on 895 rows (bar 98.5%). Weights are public on the Hugging Face Hub (fastino-community license); the lab scored revision a7e4589. Decision Index 0.4.1 (release-v4), source generated 2026-10-29T14:04:16Z. Full = 60% equated private same-skill + 40% equated private new-domain tasks; public contributes zero. Both private components are mapped onto the public scale using untuned stock models. Public weighting/panel changed (GSM8K rebuilt, ForecastBench retired, WinoGrande area reassigned); these are new benchmark definitions, not proof of a model upgrade. Public 37-benchmark index and 42 published native benchmarks remain separate. Preserve source rounding, coverage, exclusions and unknown execution dates; no new uncertainty inferred. Source notes: New in the 2026-10-21 PR batch (PR #255). Replaces the earlier board entry decision-27b-ckpt-nothink (newer version of the same model). Submitted public 0.3 run adopted (kit rescore 71.99). Private tests run by the lab on one RTX PRO 6000 with the author's serving; decision agreement with the submitted rows 99.90% on 895 rows (bar 98.5%). Weights are public on the Hugging Face Hub (fastino-community license); the lab scored revision a7e4589. Author public-run adoption/execution status and private lab setup are disclosed in configuration; evaluator-published does not mean every public run was independently executed. 32 benchmarks with a per-field right or wrong answer (ToolRet, BRIGHT and RouterBench excluded), each weighted equally; a 1-in-6 sample of requests; confidence is the probability placed on the chosen option. Rows an entrant trained on count as wrong here too. Contributing benchmarks 31. The board served a7e45896be; the registry pins the newest repository revision, and the commit titles after it concern the card. Decision Index 0.4 (release-v4), source generated 2026-10-20T14:13:56Z. Full = 60% equated private same-skill + 40% equated private new-domain tasks; public contributes zero. Both private components are mapped onto the public scale using untuned stock models. Public weighting/panel changed (GSM8K rebuilt, ForecastBench retired, WinoGrande area reassigned); these are new benchmark definitions, not proof of a model upgrade. Public 37-benchmark index and 42 published native benchmarks remain separate. Preserve source rounding, coverage, exclusions and unknown execution dates; no new uncertainty inferred. Checked / reported: 2026-11-02 · Samples: 34411 Snapshot: 2026-10-20T14:13:56Z · Revision: 133313cf8963f2b6bed960abb0152fd14cbb6731; c130faf65f82e81d3bdefb01e758c712dc5c5199 for the 2 November 2026 additions · Published observations; result reuse terms not specified Result source · Benchmark source
HF text calibration 0.4 · Brier	0.2568 Brier	Setup & source HF text calibration 0.4 · Brier Exact source engine glide-v1.2; {"kind": "full fine-tune", "technique": "autoregressive", "technique_name": "autoregressive", "served_params": 27800000000, "params_estimated": false, "params_from_name": false, "trained_bytes": null, "model_url": "https://huggingface.co/fastino/GLiDE", "code_url": "https://huggingface.co/fastino/GLiDE/tree/main/glide", "base_model": "Qwen/Qwen3.8-27B", "base_url": "https://huggingface.co/Qwen/Qwen3.8-27B", "weights_repo": "fastino/GLiDE", "served_checkpoint": "a7e45896be7c96ba3354891da790c5ec99c30b6c", "weights_note": null, "org": {"id": "fastino", "name": "Fastino", "logo": "data/orgs/fastino.png"}, "pr": 255, "replaces": "decision-27b-ckpt-nothink", "new": true}. New in the 2026-10-21 PR batch (PR #255). Replaces the earlier board entry decision-27b-ckpt-nothink (newer version of the same model). Submitted public 0.3 run adopted (kit rescore 71.99). Private tests run by the lab on one RTX PRO 6000 with the author's serving; decision agreement with the submitted rows 99.90% on 895 rows (bar 98.5%). Weights are public on the Hugging Face Hub (fastino-community license); the lab scored revision a7e4589. Decision Index 0.4.1 (release-v4), source generated 2026-10-29T14:04:16Z. Full = 60% equated private same-skill + 40% equated private new-domain tasks; public contributes zero. Both private components are mapped onto the public scale using untuned stock models. Public weighting/panel changed (GSM8K rebuilt, ForecastBench retired, WinoGrande area reassigned); these are new benchmark definitions, not proof of a model upgrade. Public 37-benchmark index and 42 published native benchmarks remain separate. Preserve source rounding, coverage, exclusions and unknown execution dates; no new uncertainty inferred. Source notes: New in the 2026-10-21 PR batch (PR #255). Replaces the earlier board entry decision-27b-ckpt-nothink (newer version of the same model). Submitted public 0.3 run adopted (kit rescore 71.99). Private tests run by the lab on one RTX PRO 6000 with the author's serving; decision agreement with the submitted rows 99.90% on 895 rows (bar 98.5%). Weights are public on the Hugging Face Hub (fastino-community license); the lab scored revision a7e4589. Author public-run adoption/execution status and private lab setup are disclosed in configuration; evaluator-published does not mean every public run was independently executed. 32 benchmarks with a per-field right or wrong answer (ToolRet, BRIGHT and RouterBench excluded), each weighted equally; a 1-in-6 sample of requests; confidence is the probability placed on the chosen option. Rows an entrant trained on count as wrong here too. Contributing benchmarks 31. The board served a7e45896be; the registry pins the newest repository revision, and the commit titles after it concern the card. Decision Index 0.4 (release-v4), source generated 2026-10-20T14:13:56Z. Full = 60% equated private same-skill + 40% equated private new-domain tasks; public contributes zero. Both private components are mapped onto the public scale using untuned stock models. Public weighting/panel changed (GSM8K rebuilt, ForecastBench retired, WinoGrande area reassigned); these are new benchmark definitions, not proof of a model upgrade. Public 37-benchmark index and 42 published native benchmarks remain separate. Preserve source rounding, coverage, exclusions and unknown execution dates; no new uncertainty inferred. Checked / reported: 2026-11-02 · Samples: 34411 Snapshot: 2026-10-20T14:13:56Z · Revision: 133313cf8963f2b6bed960abb0152fd14cbb6731; c130faf65f82e81d3bdefb01e758c712dc5c5199 for the 2 November 2026 additions · Published observations; result reuse terms not specified Result source · Benchmark source
HF text calibration 0.4 · Sample accuracy	81.49%	Setup & source HF text calibration 0.4 · Sample accuracy Exact source engine glide-v1.2; {"kind": "full fine-tune", "technique": "autoregressive", "technique_name": "autoregressive", "served_params": 27800000000, "params_estimated": false, "params_from_name": false, "trained_bytes": null, "model_url": "https://huggingface.co/fastino/GLiDE", "code_url": "https://huggingface.co/fastino/GLiDE/tree/main/glide", "base_model": "Qwen/Qwen3.8-27B", "base_url": "https://huggingface.co/Qwen/Qwen3.8-27B", "weights_repo": "fastino/GLiDE", "served_checkpoint": "a7e45896be7c96ba3354891da790c5ec99c30b6c", "weights_note": null, "org": {"id": "fastino", "name": "Fastino", "logo": "data/orgs/fastino.png"}, "pr": 255, "replaces": "decision-27b-ckpt-nothink", "new": true}. New in the 2026-10-21 PR batch (PR #255). Replaces the earlier board entry decision-27b-ckpt-nothink (newer version of the same model). Submitted public 0.3 run adopted (kit rescore 71.99). Private tests run by the lab on one RTX PRO 6000 with the author's serving; decision agreement with the submitted rows 99.90% on 895 rows (bar 98.5%). Weights are public on the Hugging Face Hub (fastino-community license); the lab scored revision a7e4589. Decision Index 0.4.1 (release-v4), source generated 2026-10-29T14:04:16Z. Full = 60% equated private same-skill + 40% equated private new-domain tasks; public contributes zero. Both private components are mapped onto the public scale using untuned stock models. Public weighting/panel changed (GSM8K rebuilt, ForecastBench retired, WinoGrande area reassigned); these are new benchmark definitions, not proof of a model upgrade. Public 37-benchmark index and 42 published native benchmarks remain separate. Preserve source rounding, coverage, exclusions and unknown execution dates; no new uncertainty inferred. Source notes: New in the 2026-10-21 PR batch (PR #255). Replaces the earlier board entry decision-27b-ckpt-nothink (newer version of the same model). Submitted public 0.3 run adopted (kit rescore 71.99). Private tests run by the lab on one RTX PRO 6000 with the author's serving; decision agreement with the submitted rows 99.90% on 895 rows (bar 98.5%). Weights are public on the Hugging Face Hub (fastino-community license); the lab scored revision a7e4589. Author public-run adoption/execution status and private lab setup are disclosed in configuration; evaluator-published does not mean every public run was independently executed. 32 benchmarks with a per-field right or wrong answer (ToolRet, BRIGHT and RouterBench excluded), each weighted equally; a 1-in-6 sample of requests; confidence is the probability placed on the chosen option. Rows an entrant trained on count as wrong here too. Contributing benchmarks 31. The board served a7e45896be; the registry pins the newest repository revision, and the commit titles after it concern the card. Decision Index 0.4 (release-v4), source generated 2026-10-20T14:13:56Z. Full = 60% equated private same-skill + 40% equated private new-domain tasks; public contributes zero. Both private components are mapped onto the public scale using untuned stock models. Public weighting/panel changed (GSM8K rebuilt, ForecastBench retired, WinoGrande area reassigned); these are new benchmark definitions, not proof of a model upgrade. Public 37-benchmark index and 42 published native benchmarks remain separate. Preserve source rounding, coverage, exclusions and unknown execution dates; no new uncertainty inferred. Checked / reported: 2026-11-02 · Samples: 34411 Snapshot: 2026-10-20T14:13:56Z · Revision: 133313cf8963f2b6bed960abb0152fd14cbb6731; c130faf65f82e81d3bdefb01e758c712dc5c5199 for the 2 November 2026 additions · Published observations; result reuse terms not specified Result source · Benchmark source
HF text calibration 0.4 · Mean chosen confidence	82.62%	Setup & source HF text calibration 0.4 · Mean chosen confidence Exact source engine glide-v1.2; {"kind": "full fine-tune", "technique": "autoregressive", "technique_name": "autoregressive", "served_params": 27800000000, "params_estimated": false, "params_from_name": false, "trained_bytes": null, "model_url": "https://huggingface.co/fastino/GLiDE", "code_url": "https://huggingface.co/fastino/GLiDE/tree/main/glide", "base_model": "Qwen/Qwen3.8-27B", "base_url": "https://huggingface.co/Qwen/Qwen3.8-27B", "weights_repo": "fastino/GLiDE", "served_checkpoint": "a7e45896be7c96ba3354891da790c5ec99c30b6c", "weights_note": null, "org": {"id": "fastino", "name": "Fastino", "logo": "data/orgs/fastino.png"}, "pr": 255, "replaces": "decision-27b-ckpt-nothink", "new": true}. New in the 2026-10-21 PR batch (PR #255). Replaces the earlier board entry decision-27b-ckpt-nothink (newer version of the same model). Submitted public 0.3 run adopted (kit rescore 71.99). Private tests run by the lab on one RTX PRO 6000 with the author's serving; decision agreement with the submitted rows 99.90% on 895 rows (bar 98.5%). Weights are public on the Hugging Face Hub (fastino-community license); the lab scored revision a7e4589. Decision Index 0.4.1 (release-v4), source generated 2026-10-29T14:04:16Z. Full = 60% equated private same-skill + 40% equated private new-domain tasks; public contributes zero. Both private components are mapped onto the public scale using untuned stock models. Public weighting/panel changed (GSM8K rebuilt, ForecastBench retired, WinoGrande area reassigned); these are new benchmark definitions, not proof of a model upgrade. Public 37-benchmark index and 42 published native benchmarks remain separate. Preserve source rounding, coverage, exclusions and unknown execution dates; no new uncertainty inferred. Source notes: New in the 2026-10-21 PR batch (PR #255). Replaces the earlier board entry decision-27b-ckpt-nothink (newer version of the same model). Submitted public 0.3 run adopted (kit rescore 71.99). Private tests run by the lab on one RTX PRO 6000 with the author's serving; decision agreement with the submitted rows 99.90% on 895 rows (bar 98.5%). Weights are public on the Hugging Face Hub (fastino-community license); the lab scored revision a7e4589. Author public-run adoption/execution status and private lab setup are disclosed in configuration; evaluator-published does not mean every public run was independently executed. 32 benchmarks with a per-field right or wrong answer (ToolRet, BRIGHT and RouterBench excluded), each weighted equally; a 1-in-6 sample of requests; confidence is the probability placed on the chosen option. Rows an entrant trained on count as wrong here too. Contributing benchmarks 31. The board served a7e45896be; the registry pins the newest repository revision, and the commit titles after it concern the card. Decision Index 0.4 (release-v4), source generated 2026-10-20T14:13:56Z. Full = 60% equated private same-skill + 40% equated private new-domain tasks; public contributes zero. Both private components are mapped onto the public scale using untuned stock models. Public weighting/panel changed (GSM8K rebuilt, ForecastBench retired, WinoGrande area reassigned); these are new benchmark definitions, not proof of a model upgrade. Public 37-benchmark index and 42 published native benchmarks remain separate. Preserve source rounding, coverage, exclusions and unknown execution dates; no new uncertainty inferred. Checked / reported: 2026-11-02 · Samples: 34411 Snapshot: 2026-10-20T14:13:56Z · Revision: 133313cf8963f2b6bed960abb0152fd14cbb6731; c130faf65f82e81d3bdefb01e758c712dc5c5199 for the 2 November 2026 additions · Published observations; result reuse terms not specified Result source · Benchmark source
HF text calibration 0.4 · Wrong among answers at ≥95% confidence	0.72%	Setup & source HF text calibration 0.4 · Wrong among answers at ≥95% confidence Exact source engine glide-v1.2; {"kind": "full fine-tune", "technique": "autoregressive", "technique_name": "autoregressive", "served_params": 27800000000, "params_estimated": false, "params_from_name": false, "trained_bytes": null, "model_url": "https://huggingface.co/fastino/GLiDE", "code_url": "https://huggingface.co/fastino/GLiDE/tree/main/glide", "base_model": "Qwen/Qwen3.8-27B", "base_url": "https://huggingface.co/Qwen/Qwen3.8-27B", "weights_repo": "fastino/GLiDE", "served_checkpoint": "a7e45896be7c96ba3354891da790c5ec99c30b6c", "weights_note": null, "org": {"id": "fastino", "name": "Fastino", "logo": "data/orgs/fastino.png"}, "pr": 255, "replaces": "decision-27b-ckpt-nothink", "new": true}. New in the 2026-10-21 PR batch (PR #255). Replaces the earlier board entry decision-27b-ckpt-nothink (newer version of the same model). Submitted public 0.3 run adopted (kit rescore 71.99). Private tests run by the lab on one RTX PRO 6000 with the author's serving; decision agreement with the submitted rows 99.90% on 895 rows (bar 98.5%). Weights are public on the Hugging Face Hub (fastino-community license); the lab scored revision a7e4589. Decision Index 0.4.1 (release-v4), source generated 2026-10-29T14:04:16Z. Full = 60% equated private same-skill + 40% equated private new-domain tasks; public contributes zero. Both private components are mapped onto the public scale using untuned stock models. Public weighting/panel changed (GSM8K rebuilt, ForecastBench retired, WinoGrande area reassigned); these are new benchmark definitions, not proof of a model upgrade. Public 37-benchmark index and 42 published native benchmarks remain separate. Preserve source rounding, coverage, exclusions and unknown execution dates; no new uncertainty inferred. Source notes: New in the 2026-10-21 PR batch (PR #255). Replaces the earlier board entry decision-27b-ckpt-nothink (newer version of the same model). Submitted public 0.3 run adopted (kit rescore 71.99). Private tests run by the lab on one RTX PRO 6000 with the author's serving; decision agreement with the submitted rows 99.90% on 895 rows (bar 98.5%). Weights are public on the Hugging Face Hub (fastino-community license); the lab scored revision a7e4589. Author public-run adoption/execution status and private lab setup are disclosed in configuration; evaluator-published does not mean every public run was independently executed. 32 benchmarks with a per-field right or wrong answer (ToolRet, BRIGHT and RouterBench excluded), each weighted equally; a 1-in-6 sample of requests; confidence is the probability placed on the chosen option. Rows an entrant trained on count as wrong here too. Contributing benchmarks 31. The board served a7e45896be; the registry pins the newest repository revision, and the commit titles after it concern the card. Decision Index 0.4 (release-v4), source generated 2026-10-20T14:13:56Z. Full = 60% equated private same-skill + 40% equated private new-domain tasks; public contributes zero. Both private components are mapped onto the public scale using untuned stock models. Public weighting/panel changed (GSM8K rebuilt, ForecastBench retired, WinoGrande area reassigned); these are new benchmark definitions, not proof of a model upgrade. Public 37-benchmark index and 42 published native benchmarks remain separate. Preserve source rounding, coverage, exclusions and unknown execution dates; no new uncertainty inferred. Checked / reported: 2026-11-02 · Samples: 34411 Snapshot: 2026-10-20T14:13:56Z · Revision: 133313cf8963f2b6bed960abb0152fd14cbb6731; c130faf65f82e81d3bdefb01e758c712dc5c5199 for the 2 November 2026 additions · Published observations; result reuse terms not specified Result source · Benchmark source
HF text calibration 0.4 · Knowledge · ECE	7.53 pp	Setup & source HF text calibration 0.4 · Knowledge · ECE Exact source engine glide-v1.2; {"kind": "full fine-tune", "technique": "autoregressive", "technique_name": "autoregressive", "served_params": 27800000000, "params_estimated": false, "params_from_name": false, "trained_bytes": null, "model_url": "https://huggingface.co/fastino/GLiDE", "code_url": "https://huggingface.co/fastino/GLiDE/tree/main/glide", "base_model": "Qwen/Qwen3.8-27B", "base_url": "https://huggingface.co/Qwen/Qwen3.8-27B", "weights_repo": "fastino/GLiDE", "served_checkpoint": "a7e45896be7c96ba3354891da790c5ec99c30b6c", "weights_note": null, "org": {"id": "fastino", "name": "Fastino", "logo": "data/orgs/fastino.png"}, "pr": 255, "replaces": "decision-27b-ckpt-nothink", "new": true}. New in the 2026-10-21 PR batch (PR #255). Replaces the earlier board entry decision-27b-ckpt-nothink (newer version of the same model). Submitted public 0.3 run adopted (kit rescore 71.99). Private tests run by the lab on one RTX PRO 6000 with the author's serving; decision agreement with the submitted rows 99.90% on 895 rows (bar 98.5%). Weights are public on the Hugging Face Hub (fastino-community license); the lab scored revision a7e4589. Decision Index 0.4.1 (release-v4), source generated 2026-10-29T14:04:16Z. Full = 60% equated private same-skill + 40% equated private new-domain tasks; public contributes zero. Both private components are mapped onto the public scale using untuned stock models. Public weighting/panel changed (GSM8K rebuilt, ForecastBench retired, WinoGrande area reassigned); these are new benchmark definitions, not proof of a model upgrade. Public 37-benchmark index and 42 published native benchmarks remain separate. Preserve source rounding, coverage, exclusions and unknown execution dates; no new uncertainty inferred. Source notes: New in the 2026-10-21 PR batch (PR #255). Replaces the earlier board entry decision-27b-ckpt-nothink (newer version of the same model). Submitted public 0.3 run adopted (kit rescore 71.99). Private tests run by the lab on one RTX PRO 6000 with the author's serving; decision agreement with the submitted rows 99.90% on 895 rows (bar 98.5%). Weights are public on the Hugging Face Hub (fastino-community license); the lab scored revision a7e4589. Author public-run adoption/execution status and private lab setup are disclosed in configuration; evaluator-published does not mean every public run was independently executed. Area-specific item counts unpublished; do not use whole-sample count. The board served a7e45896be; the registry pins the newest repository revision, and the commit titles after it concern the card. Decision Index 0.4 (release-v4), source generated 2026-10-20T14:13:56Z. Full = 60% equated private same-skill + 40% equated private new-domain tasks; public contributes zero. Both private components are mapped onto the public scale using untuned stock models. Public weighting/panel changed (GSM8K rebuilt, ForecastBench retired, WinoGrande area reassigned); these are new benchmark definitions, not proof of a model upgrade. Public 37-benchmark index and 42 published native benchmarks remain separate. Preserve source rounding, coverage, exclusions and unknown execution dates; no new uncertainty inferred. Checked / reported: 2026-11-02 · Samples: Not published Snapshot: 2026-10-20T14:13:56Z · Revision: 133313cf8963f2b6bed960abb0152fd14cbb6731; c130faf65f82e81d3bdefb01e758c712dc5c5199 for the 2 November 2026 additions · Published observations; result reuse terms not specified Result source · Benchmark source
HF text calibration 0.4 · Knowledge · Sample accuracy	72.87%	Setup & source HF text calibration 0.4 · Knowledge · Sample accuracy Exact source engine glide-v1.2; {"kind": "full fine-tune", "technique": "autoregressive", "technique_name": "autoregressive", "served_params": 27800000000, "params_estimated": false, "params_from_name": false, "trained_bytes": null, "model_url": "https://huggingface.co/fastino/GLiDE", "code_url": "https://huggingface.co/fastino/GLiDE/tree/main/glide", "base_model": "Qwen/Qwen3.8-27B", "base_url": "https://huggingface.co/Qwen/Qwen3.8-27B", "weights_repo": "fastino/GLiDE", "served_checkpoint": "a7e45896be7c96ba3354891da790c5ec99c30b6c", "weights_note": null, "org": {"id": "fastino", "name": "Fastino", "logo": "data/orgs/fastino.png"}, "pr": 255, "replaces": "decision-27b-ckpt-nothink", "new": true}. New in the 2026-10-21 PR batch (PR #255). Replaces the earlier board entry decision-27b-ckpt-nothink (newer version of the same model). Submitted public 0.3 run adopted (kit rescore 71.99). Private tests run by the lab on one RTX PRO 6000 with the author's serving; decision agreement with the submitted rows 99.90% on 895 rows (bar 98.5%). Weights are public on the Hugging Face Hub (fastino-community license); the lab scored revision a7e4589. Decision Index 0.4.1 (release-v4), source generated 2026-10-29T14:04:16Z. Full = 60% equated private same-skill + 40% equated private new-domain tasks; public contributes zero. Both private components are mapped onto the public scale using untuned stock models. Public weighting/panel changed (GSM8K rebuilt, ForecastBench retired, WinoGrande area reassigned); these are new benchmark definitions, not proof of a model upgrade. Public 37-benchmark index and 42 published native benchmarks remain separate. Preserve source rounding, coverage, exclusions and unknown execution dates; no new uncertainty inferred. Source notes: New in the 2026-10-21 PR batch (PR #255). Replaces the earlier board entry decision-27b-ckpt-nothink (newer version of the same model). Submitted public 0.3 run adopted (kit rescore 71.99). Private tests run by the lab on one RTX PRO 6000 with the author's serving; decision agreement with the submitted rows 99.90% on 895 rows (bar 98.5%). Weights are public on the Hugging Face Hub (fastino-community license); the lab scored revision a7e4589. Author public-run adoption/execution status and private lab setup are disclosed in configuration; evaluator-published does not mean every public run was independently executed. Area-specific item counts unpublished; do not use whole-sample count. The board served a7e45896be; the registry pins the newest repository revision, and the commit titles after it concern the card. Decision Index 0.4 (release-v4), source generated 2026-10-20T14:13:56Z. Full = 60% equated private same-skill + 40% equated private new-domain tasks; public contributes zero. Both private components are mapped onto the public scale using untuned stock models. Public weighting/panel changed (GSM8K rebuilt, ForecastBench retired, WinoGrande area reassigned); these are new benchmark definitions, not proof of a model upgrade. Public 37-benchmark index and 42 published native benchmarks remain separate. Preserve source rounding, coverage, exclusions and unknown execution dates; no new uncertainty inferred. Checked / reported: 2026-11-02 · Samples: Not published Snapshot: 2026-10-20T14:13:56Z · Revision: 133313cf8963f2b6bed960abb0152fd14cbb6731; c130faf65f82e81d3bdefb01e758c712dc5c5199 for the 2 November 2026 additions · Published observations; result reuse terms not specified Result source · Benchmark source
HF text calibration 0.4 · Knowledge · Mean chosen confidence	77.53%	Setup & source HF text calibration 0.4 · Knowledge · Mean chosen confidence Exact source engine glide-v1.2; {"kind": "full fine-tune", "technique": "autoregressive", "technique_name": "autoregressive", "served_params": 27800000000, "params_estimated": false, "params_from_name": false, "trained_bytes": null, "model_url": "https://huggingface.co/fastino/GLiDE", "code_url": "https://huggingface.co/fastino/GLiDE/tree/main/glide", "base_model": "Qwen/Qwen3.8-27B", "base_url": "https://huggingface.co/Qwen/Qwen3.8-27B", "weights_repo": "fastino/GLiDE", "served_checkpoint": "a7e45896be7c96ba3354891da790c5ec99c30b6c", "weights_note": null, "org": {"id": "fastino", "name": "Fastino", "logo": "data/orgs/fastino.png"}, "pr": 255, "replaces": "decision-27b-ckpt-nothink", "new": true}. New in the 2026-10-21 PR batch (PR #255). Replaces the earlier board entry decision-27b-ckpt-nothink (newer version of the same model). Submitted public 0.3 run adopted (kit rescore 71.99). Private tests run by the lab on one RTX PRO 6000 with the author's serving; decision agreement with the submitted rows 99.90% on 895 rows (bar 98.5%). Weights are public on the Hugging Face Hub (fastino-community license); the lab scored revision a7e4589. Decision Index 0.4.1 (release-v4), source generated 2026-10-29T14:04:16Z. Full = 60% equated private same-skill + 40% equated private new-domain tasks; public contributes zero. Both private components are mapped onto the public scale using untuned stock models. Public weighting/panel changed (GSM8K rebuilt, ForecastBench retired, WinoGrande area reassigned); these are new benchmark definitions, not proof of a model upgrade. Public 37-benchmark index and 42 published native benchmarks remain separate. Preserve source rounding, coverage, exclusions and unknown execution dates; no new uncertainty inferred. Source notes: New in the 2026-10-21 PR batch (PR #255). Replaces the earlier board entry decision-27b-ckpt-nothink (newer version of the same model). Submitted public 0.3 run adopted (kit rescore 71.99). Private tests run by the lab on one RTX PRO 6000 with the author's serving; decision agreement with the submitted rows 99.90% on 895 rows (bar 98.5%). Weights are public on the Hugging Face Hub (fastino-community license); the lab scored revision a7e4589. Author public-run adoption/execution status and private lab setup are disclosed in configuration; evaluator-published does not mean every public run was independently executed. Area-specific item counts unpublished; do not use whole-sample count. The board served a7e45896be; the registry pins the newest repository revision, and the commit titles after it concern the card. Decision Index 0.4 (release-v4), source generated 2026-10-20T14:13:56Z. Full = 60% equated private same-skill + 40% equated private new-domain tasks; public contributes zero. Both private components are mapped onto the public scale using untuned stock models. Public weighting/panel changed (GSM8K rebuilt, ForecastBench retired, WinoGrande area reassigned); these are new benchmark definitions, not proof of a model upgrade. Public 37-benchmark index and 42 published native benchmarks remain separate. Preserve source rounding, coverage, exclusions and unknown execution dates; no new uncertainty inferred. Checked / reported: 2026-11-02 · Samples: Not published Snapshot: 2026-10-20T14:13:56Z · Revision: 133313cf8963f2b6bed960abb0152fd14cbb6731; c130faf65f82e81d3bdefb01e758c712dc5c5199 for the 2 November 2026 additions · Published observations; result reuse terms not specified Result source · Benchmark source
HF text calibration 0.4 · Language · ECE	1.81 pp	Setup & source HF text calibration 0.4 · Language · ECE Exact source engine glide-v1.2; {"kind": "full fine-tune", "technique": "autoregressive", "technique_name": "autoregressive", "served_params": 27800000000, "params_estimated": false, "params_from_name": false, "trained_bytes": null, "model_url": "https://huggingface.co/fastino/GLiDE", "code_url": "https://huggingface.co/fastino/GLiDE/tree/main/glide", "base_model": "Qwen/Qwen3.8-27B", "base_url": "https://huggingface.co/Qwen/Qwen3.8-27B", "weights_repo": "fastino/GLiDE", "served_checkpoint": "a7e45896be7c96ba3354891da790c5ec99c30b6c", "weights_note": null, "org": {"id": "fastino", "name": "Fastino", "logo": "data/orgs/fastino.png"}, "pr": 255, "replaces": "decision-27b-ckpt-nothink", "new": true}. New in the 2026-10-21 PR batch (PR #255). Replaces the earlier board entry decision-27b-ckpt-nothink (newer version of the same model). Submitted public 0.3 run adopted (kit rescore 71.99). Private tests run by the lab on one RTX PRO 6000 with the author's serving; decision agreement with the submitted rows 99.90% on 895 rows (bar 98.5%). Weights are public on the Hugging Face Hub (fastino-community license); the lab scored revision a7e4589. Decision Index 0.4.1 (release-v4), source generated 2026-10-29T14:04:16Z. Full = 60% equated private same-skill + 40% equated private new-domain tasks; public contributes zero. Both private components are mapped onto the public scale using untuned stock models. Public weighting/panel changed (GSM8K rebuilt, ForecastBench retired, WinoGrande area reassigned); these are new benchmark definitions, not proof of a model upgrade. Public 37-benchmark index and 42 published native benchmarks remain separate. Preserve source rounding, coverage, exclusions and unknown execution dates; no new uncertainty inferred. Source notes: New in the 2026-10-21 PR batch (PR #255). Replaces the earlier board entry decision-27b-ckpt-nothink (newer version of the same model). Submitted public 0.3 run adopted (kit rescore 71.99). Private tests run by the lab on one RTX PRO 6000 with the author's serving; decision agreement with the submitted rows 99.90% on 895 rows (bar 98.5%). Weights are public on the Hugging Face Hub (fastino-community license); the lab scored revision a7e4589. Author public-run adoption/execution status and private lab setup are disclosed in configuration; evaluator-published does not mean every public run was independently executed. Area-specific item counts unpublished; do not use whole-sample count. The board served a7e45896be; the registry pins the newest repository revision, and the commit titles after it concern the card. Decision Index 0.4 (release-v4), source generated 2026-10-20T14:13:56Z. Full = 60% equated private same-skill + 40% equated private new-domain tasks; public contributes zero. Both private components are mapped onto the public scale using untuned stock models. Public weighting/panel changed (GSM8K rebuilt, ForecastBench retired, WinoGrande area reassigned); these are new benchmark definitions, not proof of a model upgrade. Public 37-benchmark index and 42 published native benchmarks remain separate. Preserve source rounding, coverage, exclusions and unknown execution dates; no new uncertainty inferred. Checked / reported: 2026-11-02 · Samples: Not published Snapshot: 2026-10-20T14:13:56Z · Revision: 133313cf8963f2b6bed960abb0152fd14cbb6731; c130faf65f82e81d3bdefb01e758c712dc5c5199 for the 2 November 2026 additions · Published observations; result reuse terms not specified Result source · Benchmark source
HF text calibration 0.4 · Language · Sample accuracy	90.15%	Setup & source HF text calibration 0.4 · Language · Sample accuracy Exact source engine glide-v1.2; {"kind": "full fine-tune", "technique": "autoregressive", "technique_name": "autoregressive", "served_params": 27800000000, "params_estimated": false, "params_from_name": false, "trained_bytes": null, "model_url": "https://huggingface.co/fastino/GLiDE", "code_url": "https://huggingface.co/fastino/GLiDE/tree/main/glide", "base_model": "Qwen/Qwen3.8-27B", "base_url": "https://huggingface.co/Qwen/Qwen3.8-27B", "weights_repo": "fastino/GLiDE", "served_checkpoint": "a7e45896be7c96ba3354891da790c5ec99c30b6c", "weights_note": null, "org": {"id": "fastino", "name": "Fastino", "logo": "data/orgs/fastino.png"}, "pr": 255, "replaces": "decision-27b-ckpt-nothink", "new": true}. New in the 2026-10-21 PR batch (PR #255). Replaces the earlier board entry decision-27b-ckpt-nothink (newer version of the same model). Submitted public 0.3 run adopted (kit rescore 71.99). Private tests run by the lab on one RTX PRO 6000 with the author's serving; decision agreement with the submitted rows 99.90% on 895 rows (bar 98.5%). Weights are public on the Hugging Face Hub (fastino-community license); the lab scored revision a7e4589. Decision Index 0.4.1 (release-v4), source generated 2026-10-29T14:04:16Z. Full = 60% equated private same-skill + 40% equated private new-domain tasks; public contributes zero. Both private components are mapped onto the public scale using untuned stock models. Public weighting/panel changed (GSM8K rebuilt, ForecastBench retired, WinoGrande area reassigned); these are new benchmark definitions, not proof of a model upgrade. Public 37-benchmark index and 42 published native benchmarks remain separate. Preserve source rounding, coverage, exclusions and unknown execution dates; no new uncertainty inferred. Source notes: New in the 2026-10-21 PR batch (PR #255). Replaces the earlier board entry decision-27b-ckpt-nothink (newer version of the same model). Submitted public 0.3 run adopted (kit rescore 71.99). Private tests run by the lab on one RTX PRO 6000 with the author's serving; decision agreement with the submitted rows 99.90% on 895 rows (bar 98.5%). Weights are public on the Hugging Face Hub (fastino-community license); the lab scored revision a7e4589. Author public-run adoption/execution status and private lab setup are disclosed in configuration; evaluator-published does not mean every public run was independently executed. Area-specific item counts unpublished; do not use whole-sample count. The board served a7e45896be; the registry pins the newest repository revision, and the commit titles after it concern the card. Decision Index 0.4 (release-v4), source generated 2026-10-20T14:13:56Z. Full = 60% equated private same-skill + 40% equated private new-domain tasks; public contributes zero. Both private components are mapped onto the public scale using untuned stock models. Public weighting/panel changed (GSM8K rebuilt, ForecastBench retired, WinoGrande area reassigned); these are new benchmark definitions, not proof of a model upgrade. Public 37-benchmark index and 42 published native benchmarks remain separate. Preserve source rounding, coverage, exclusions and unknown execution dates; no new uncertainty inferred. Checked / reported: 2026-11-02 · Samples: Not published Snapshot: 2026-10-20T14:13:56Z · Revision: 133313cf8963f2b6bed960abb0152fd14cbb6731; c130faf65f82e81d3bdefb01e758c712dc5c5199 for the 2 November 2026 additions · Published observations; result reuse terms not specified Result source · Benchmark source
HF text calibration 0.4 · Language · Mean chosen confidence	88.34%	Setup & source HF text calibration 0.4 · Language · Mean chosen confidence Exact source engine glide-v1.2; {"kind": "full fine-tune", "technique": "autoregressive", "technique_name": "autoregressive", "served_params": 27800000000, "params_estimated": false, "params_from_name": false, "trained_bytes": null, "model_url": "https://huggingface.co/fastino/GLiDE", "code_url": "https://huggingface.co/fastino/GLiDE/tree/main/glide", "base_model": "Qwen/Qwen3.8-27B", "base_url": "https://huggingface.co/Qwen/Qwen3.8-27B", "weights_repo": "fastino/GLiDE", "served_checkpoint": "a7e45896be7c96ba3354891da790c5ec99c30b6c", "weights_note": null, "org": {"id": "fastino", "name": "Fastino", "logo": "data/orgs/fastino.png"}, "pr": 255, "replaces": "decision-27b-ckpt-nothink", "new": true}. New in the 2026-10-21 PR batch (PR #255). Replaces the earlier board entry decision-27b-ckpt-nothink (newer version of the same model). Submitted public 0.3 run adopted (kit rescore 71.99). Private tests run by the lab on one RTX PRO 6000 with the author's serving; decision agreement with the submitted rows 99.90% on 895 rows (bar 98.5%). Weights are public on the Hugging Face Hub (fastino-community license); the lab scored revision a7e4589. Decision Index 0.4.1 (release-v4), source generated 2026-10-29T14:04:16Z. Full = 60% equated private same-skill + 40% equated private new-domain tasks; public contributes zero. Both private components are mapped onto the public scale using untuned stock models. Public weighting/panel changed (GSM8K rebuilt, ForecastBench retired, WinoGrande area reassigned); these are new benchmark definitions, not proof of a model upgrade. Public 37-benchmark index and 42 published native benchmarks remain separate. Preserve source rounding, coverage, exclusions and unknown execution dates; no new uncertainty inferred. Source notes: New in the 2026-10-21 PR batch (PR #255). Replaces the earlier board entry decision-27b-ckpt-nothink (newer version of the same model). Submitted public 0.3 run adopted (kit rescore 71.99). Private tests run by the lab on one RTX PRO 6000 with the author's serving; decision agreement with the submitted rows 99.90% on 895 rows (bar 98.5%). Weights are public on the Hugging Face Hub (fastino-community license); the lab scored revision a7e4589. Author public-run adoption/execution status and private lab setup are disclosed in configuration; evaluator-published does not mean every public run was independently executed. Area-specific item counts unpublished; do not use whole-sample count. The board served a7e45896be; the registry pins the newest repository revision, and the commit titles after it concern the card. Decision Index 0.4 (release-v4), source generated 2026-10-20T14:13:56Z. Full = 60% equated private same-skill + 40% equated private new-domain tasks; public contributes zero. Both private components are mapped onto the public scale using untuned stock models. Public weighting/panel changed (GSM8K rebuilt, ForecastBench retired, WinoGrande area reassigned); these are new benchmark definitions, not proof of a model upgrade. Public 37-benchmark index and 42 published native benchmarks remain separate. Preserve source rounding, coverage, exclusions and unknown execution dates; no new uncertainty inferred. Checked / reported: 2026-11-02 · Samples: Not published Snapshot: 2026-10-20T14:13:56Z · Revision: 133313cf8963f2b6bed960abb0152fd14cbb6731; c130faf65f82e81d3bdefb01e758c712dc5c5199 for the 2 November 2026 additions · Published observations; result reuse terms not specified Result source · Benchmark source
HF text calibration 0.4 · Retrieval · ECE	3.48 pp	Setup & source HF text calibration 0.4 · Retrieval · ECE Exact source engine glide-v1.2; {"kind": "full fine-tune", "technique": "autoregressive", "technique_name": "autoregressive", "served_params": 27800000000, "params_estimated": false, "params_from_name": false, "trained_bytes": null, "model_url": "https://huggingface.co/fastino/GLiDE", "code_url": "https://huggingface.co/fastino/GLiDE/tree/main/glide", "base_model": "Qwen/Qwen3.8-27B", "base_url": "https://huggingface.co/Qwen/Qwen3.8-27B", "weights_repo": "fastino/GLiDE", "served_checkpoint": "a7e45896be7c96ba3354891da790c5ec99c30b6c", "weights_note": null, "org": {"id": "fastino", "name": "Fastino", "logo": "data/orgs/fastino.png"}, "pr": 255, "replaces": "decision-27b-ckpt-nothink", "new": true}. New in the 2026-10-21 PR batch (PR #255). Replaces the earlier board entry decision-27b-ckpt-nothink (newer version of the same model). Submitted public 0.3 run adopted (kit rescore 71.99). Private tests run by the lab on one RTX PRO 6000 with the author's serving; decision agreement with the submitted rows 99.90% on 895 rows (bar 98.5%). Weights are public on the Hugging Face Hub (fastino-community license); the lab scored revision a7e4589. Decision Index 0.4.1 (release-v4), source generated 2026-10-29T14:04:16Z. Full = 60% equated private same-skill + 40% equated private new-domain tasks; public contributes zero. Both private components are mapped onto the public scale using untuned stock models. Public weighting/panel changed (GSM8K rebuilt, ForecastBench retired, WinoGrande area reassigned); these are new benchmark definitions, not proof of a model upgrade. Public 37-benchmark index and 42 published native benchmarks remain separate. Preserve source rounding, coverage, exclusions and unknown execution dates; no new uncertainty inferred. Source notes: New in the 2026-10-21 PR batch (PR #255). Replaces the earlier board entry decision-27b-ckpt-nothink (newer version of the same model). Submitted public 0.3 run adopted (kit rescore 71.99). Private tests run by the lab on one RTX PRO 6000 with the author's serving; decision agreement with the submitted rows 99.90% on 895 rows (bar 98.5%). Weights are public on the Hugging Face Hub (fastino-community license); the lab scored revision a7e4589. Author public-run adoption/execution status and private lab setup are disclosed in configuration; evaluator-published does not mean every public run was independently executed. Area-specific item counts unpublished; do not use whole-sample count. The board served a7e45896be; the registry pins the newest repository revision, and the commit titles after it concern the card. Decision Index 0.4 (release-v4), source generated 2026-10-20T14:13:56Z. Full = 60% equated private same-skill + 40% equated private new-domain tasks; public contributes zero. Both private components are mapped onto the public scale using untuned stock models. Public weighting/panel changed (GSM8K rebuilt, ForecastBench retired, WinoGrande area reassigned); these are new benchmark definitions, not proof of a model upgrade. Public 37-benchmark index and 42 published native benchmarks remain separate. Preserve source rounding, coverage, exclusions and unknown execution dates; no new uncertainty inferred. Checked / reported: 2026-11-02 · Samples: Not published Snapshot: 2026-10-20T14:13:56Z · Revision: 133313cf8963f2b6bed960abb0152fd14cbb6731; c130faf65f82e81d3bdefb01e758c712dc5c5199 for the 2 November 2026 additions · Published observations; result reuse terms not specified Result source · Benchmark source
HF text calibration 0.4 · Retrieval · Sample accuracy	81.63%	Setup & source HF text calibration 0.4 · Retrieval · Sample accuracy Exact source engine glide-v1.2; {"kind": "full fine-tune", "technique": "autoregressive", "technique_name": "autoregressive", "served_params": 27800000000, "params_estimated": false, "params_from_name": false, "trained_bytes": null, "model_url": "https://huggingface.co/fastino/GLiDE", "code_url": "https://huggingface.co/fastino/GLiDE/tree/main/glide", "base_model": "Qwen/Qwen3.8-27B", "base_url": "https://huggingface.co/Qwen/Qwen3.8-27B", "weights_repo": "fastino/GLiDE", "served_checkpoint": "a7e45896be7c96ba3354891da790c5ec99c30b6c", "weights_note": null, "org": {"id": "fastino", "name": "Fastino", "logo": "data/orgs/fastino.png"}, "pr": 255, "replaces": "decision-27b-ckpt-nothink", "new": true}. New in the 2026-10-21 PR batch (PR #255). Replaces the earlier board entry decision-27b-ckpt-nothink (newer version of the same model). Submitted public 0.3 run adopted (kit rescore 71.99). Private tests run by the lab on one RTX PRO 6000 with the author's serving; decision agreement with the submitted rows 99.90% on 895 rows (bar 98.5%). Weights are public on the Hugging Face Hub (fastino-community license); the lab scored revision a7e4589. Decision Index 0.4.1 (release-v4), source generated 2026-10-29T14:04:16Z. Full = 60% equated private same-skill + 40% equated private new-domain tasks; public contributes zero. Both private components are mapped onto the public scale using untuned stock models. Public weighting/panel changed (GSM8K rebuilt, ForecastBench retired, WinoGrande area reassigned); these are new benchmark definitions, not proof of a model upgrade. Public 37-benchmark index and 42 published native benchmarks remain separate. Preserve source rounding, coverage, exclusions and unknown execution dates; no new uncertainty inferred. Source notes: New in the 2026-10-21 PR batch (PR #255). Replaces the earlier board entry decision-27b-ckpt-nothink (newer version of the same model). Submitted public 0.3 run adopted (kit rescore 71.99). Private tests run by the lab on one RTX PRO 6000 with the author's serving; decision agreement with the submitted rows 99.90% on 895 rows (bar 98.5%). Weights are public on the Hugging Face Hub (fastino-community license); the lab scored revision a7e4589. Author public-run adoption/execution status and private lab setup are disclosed in configuration; evaluator-published does not mean every public run was independently executed. Area-specific item counts unpublished; do not use whole-sample count. The board served a7e45896be; the registry pins the newest repository revision, and the commit titles after it concern the card. Decision Index 0.4 (release-v4), source generated 2026-10-20T14:13:56Z. Full = 60% equated private same-skill + 40% equated private new-domain tasks; public contributes zero. Both private components are mapped onto the public scale using untuned stock models. Public weighting/panel changed (GSM8K rebuilt, ForecastBench retired, WinoGrande area reassigned); these are new benchmark definitions, not proof of a model upgrade. Public 37-benchmark index and 42 published native benchmarks remain separate. Preserve source rounding, coverage, exclusions and unknown execution dates; no new uncertainty inferred. Checked / reported: 2026-11-02 · Samples: Not published Snapshot: 2026-10-20T14:13:56Z · Revision: 133313cf8963f2b6bed960abb0152fd14cbb6731; c130faf65f82e81d3bdefb01e758c712dc5c5199 for the 2 November 2026 additions · Published observations; result reuse terms not specified Result source · Benchmark source
HF text calibration 0.4 · Retrieval · Mean chosen confidence	83.58%	Setup & source HF text calibration 0.4 · Retrieval · Mean chosen confidence Exact source engine glide-v1.2; {"kind": "full fine-tune", "technique": "autoregressive", "technique_name": "autoregressive", "served_params": 27800000000, "params_estimated": false, "params_from_name": false, "trained_bytes": null, "model_url": "https://huggingface.co/fastino/GLiDE", "code_url": "https://huggingface.co/fastino/GLiDE/tree/main/glide", "base_model": "Qwen/Qwen3.8-27B", "base_url": "https://huggingface.co/Qwen/Qwen3.8-27B", "weights_repo": "fastino/GLiDE", "served_checkpoint": "a7e45896be7c96ba3354891da790c5ec99c30b6c", "weights_note": null, "org": {"id": "fastino", "name": "Fastino", "logo": "data/orgs/fastino.png"}, "pr": 255, "replaces": "decision-27b-ckpt-nothink", "new": true}. New in the 2026-10-21 PR batch (PR #255). Replaces the earlier board entry decision-27b-ckpt-nothink (newer version of the same model). Submitted public 0.3 run adopted (kit rescore 71.99). Private tests run by the lab on one RTX PRO 6000 with the author's serving; decision agreement with the submitted rows 99.90% on 895 rows (bar 98.5%). Weights are public on the Hugging Face Hub (fastino-community license); the lab scored revision a7e4589. Decision Index 0.4.1 (release-v4), source generated 2026-10-29T14:04:16Z. Full = 60% equated private same-skill + 40% equated private new-domain tasks; public contributes zero. Both private components are mapped onto the public scale using untuned stock models. Public weighting/panel changed (GSM8K rebuilt, ForecastBench retired, WinoGrande area reassigned); these are new benchmark definitions, not proof of a model upgrade. Public 37-benchmark index and 42 published native benchmarks remain separate. Preserve source rounding, coverage, exclusions and unknown execution dates; no new uncertainty inferred. Source notes: New in the 2026-10-21 PR batch (PR #255). Replaces the earlier board entry decision-27b-ckpt-nothink (newer version of the same model). Submitted public 0.3 run adopted (kit rescore 71.99). Private tests run by the lab on one RTX PRO 6000 with the author's serving; decision agreement with the submitted rows 99.90% on 895 rows (bar 98.5%). Weights are public on the Hugging Face Hub (fastino-community license); the lab scored revision a7e4589. Author public-run adoption/execution status and private lab setup are disclosed in configuration; evaluator-published does not mean every public run was independently executed. Area-specific item counts unpublished; do not use whole-sample count. The board served a7e45896be; the registry pins the newest repository revision, and the commit titles after it concern the card. Decision Index 0.4 (release-v4), source generated 2026-10-20T14:13:56Z. Full = 60% equated private same-skill + 40% equated private new-domain tasks; public contributes zero. Both private components are mapped onto the public scale using untuned stock models. Public weighting/panel changed (GSM8K rebuilt, ForecastBench retired, WinoGrande area reassigned); these are new benchmark definitions, not proof of a model upgrade. Public 37-benchmark index and 42 published native benchmarks remain separate. Preserve source rounding, coverage, exclusions and unknown execution dates; no new uncertainty inferred. Checked / reported: 2026-11-02 · Samples: Not published Snapshot: 2026-10-20T14:13:56Z · Revision: 133313cf8963f2b6bed960abb0152fd14cbb6731; c130faf65f82e81d3bdefb01e758c712dc5c5199 for the 2 November 2026 additions · Published observations; result reuse terms not specified Result source · Benchmark source