kettei core Decision Model Results

kettei core

Explore kettei core: exact decision configuration, published specifications, source-linked benchmarks and API tariffs when available.

Published specifications

Provider: Sqwish Labs
Version: Sqwish/Kettei-Core (commit 5a98216, release tag v1.0); hosted kettei-core
System type: Trained checkpoint
Access: Open weights
License: Kettei Community Model License 1.0
Inputs: text
Decision types: Choice, Yes / no, Score
Published size: 5,058,255,360 (card); the launch post calls it 3.7B
Active parameters: Not published
Context / input limit: 4,096 input tokens by default in the reference interface (card)
Announced: 2026-10-14

Hugging Face · Announcement · Website

Model notes

Open text-only kettei decision model from Sqwish Labs, a fine-tune of IFM K2-Horizon-3.7B, announced 14 October 2026; the weights were published on 18 October under the Kettei Community Model License 1.0 with a reference package (kettei 0.1.0rc5) that scores supplied options. It is also served on OneHot as kettei-core, where fine-tuning is live. The card makes no task-quality or latency claim; the launch post’s figure values were published before the weights and do not name a revision.

Checkpoint revision: 5a9821633c8bb15b92b677d368083ec04720d164

← All decision models · Compare this decision system →

Published API tariff

$0.02 per 1M input tokens

Sqwish Labs (OneHot) · checked 2026-11-08

Launch post OneHot table, 14 October 2026: “Prices in USD per million tokens”, base inference $0.02 for kettei core. The table does not split input and output; the API bills the context it reads, so the price is stored as the input rate and output stays empty. Fine-tuned inference and training are priced separately. The post does not say whether the hosted model equals the v1.0 weights.

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

4 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 · Sqwish Labs (vendor run / launch post)

kettei core · Official / vendor · Sqwish Labs (vendor run / launch post)
Evaluation	Result	Evidence
Sqwish · kettei launch · Decision Index 0.3.1 public half · score	60.65	Setup & source Decision Index 0.3.1 public suite, scored with the maintainer’s kit on its stratified 12,000-request sample (Figure 4, Text). kettei core. Figure 4 Text point, scored with the maintainer’s kit. Published in the vendor’s launch post. Sqwish Labs’ own results in the kettei launch post of 14 October 2026. Figure values are the labelled data points the page draws (name, score, latency), read from the site bundle /static/js/main.10f99d62.js, read 8 November 2026; prose values are quoted from the same post. Values were published before the core and dot weights (18 October) and name no revision. Checked / reported: 2026-11-08 · Samples: 12000 Snapshot: 2026-10-14 · Revision: site bundle /static/js/main.10f99d62.js, read 8 November 2026 · Published observations; result reuse terms not specified Result source · Benchmark source
Sqwish · kettei launch · Decision Index 0.3.1 public half · median latency	24 ms	Setup & source Median latency per request on the same run (Figure 4, Text). kettei core. Figure 4 Text point; hardware not stated. Published in the vendor’s launch post. Sqwish Labs’ own results in the kettei launch post of 14 October 2026. Figure values are the labelled data points the page draws (name, score, latency), read from the site bundle /static/js/main.10f99d62.js, read 8 November 2026; prose values are quoted from the same post. Values were published before the core and dot weights (18 October) and name no revision. Checked / reported: 2026-11-08 · Samples: 12000 Snapshot: 2026-10-14 · Revision: site bundle /static/js/main.10f99d62.js, read 8 November 2026 · Published observations; result reuse terms not specified Result source · Benchmark source
Sqwish · kettei launch · JevBench 231 public items · accuracy	89.6%	Setup & source The 231 public JevBench items (Figure 4, JevBench). kettei core. Figure 4 JevBench point. Published in the vendor’s launch post. Sqwish Labs’ own results in the kettei launch post of 14 October 2026. Figure values are the labelled data points the page draws (name, score, latency), read from the site bundle /static/js/main.10f99d62.js, read 8 November 2026; prose values are quoted from the same post. Values were published before the core and dot weights (18 October) and name no revision. Checked / reported: 2026-11-08 · Samples: 231 Snapshot: 2026-10-14 · Revision: site bundle /static/js/main.10f99d62.js, read 8 November 2026 · Published observations; result reuse terms not specified Result source · Benchmark source
Sqwish · kettei launch · JevBench 231 public items · median end-to-end latency from the UK	147 ms	Setup & source Median end-to-end latency per request from a UK client through the provider’s public API (Figure 4, JevBench). kettei core. Figure 4 JevBench point, timed end to end from the UK. Published in the vendor’s launch post. Sqwish Labs’ own results in the kettei launch post of 14 October 2026. Figure values are the labelled data points the page draws (name, score, latency), read from the site bundle /static/js/main.10f99d62.js, read 8 November 2026; prose values are quoted from the same post. Values were published before the core and dot weights (18 October) and name no revision. Checked / reported: 2026-11-08 · Samples: 231 Snapshot: 2026-10-14 · Revision: site bundle /static/js/main.10f99d62.js, read 8 November 2026 · Published observations; result reuse terms not specified Result source · Benchmark source