H2O-Lightning-4B-Nemotron-Nano Decision Model Results

H2O-Lightning-4B-Nemotron-Nano

Explore H2O-Lightning-4B-Nemotron-Nano: exact decision configuration, published specifications, source-linked benchmarks and API tariffs when available.

Published specifications

Provider: H2O.ai
Version: v1.1
System type: Trained checkpoint
Access: Open weights
License: NVIDIA Nemotron Open Model License
Inputs: text
Decision types: Choice, Yes / no, Score
Published size: 3.97B
Active parameters: Not published
Context / input limit: 40,960 tokens as shipped; 131,072 configurable
Announced: 2026-10-09

Model card & announcement · Website

Model notes

Decision model from H2O.ai: one forward pass and one output token per decision, with a probability for every option. A fine-tune by its developer, not a third-party quantization. Announced date is the Hugging Face repository creation date; no separate announcement was found. Built on NVIDIA-Nemotron-3-Nano-4B-BF16; a separate checkpoint from H2O-Lightning-4B. Runs on unmodified vLLM 0.30.0. Card latency on one L40S over the 231 JevBench items: p50 0.028 s, p95 0.124 s, serial. System 1 v2 lists a local run of this base, which stays out because local runs are not imported. Repository created 9 October 2026; older omission surfaced with its family.

Checkpoint revision: 4e7d4ae85fd3672a023a3a1d164d29726dd15167

← All decision models · Compare this decision system →

3 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 · H2O.ai model cards (vendor measured)

H2O-Lightning-4B-Nemotron-Nano · Official / vendor · H2O.ai model cards (vendor measured)
Evaluation	Result	Evidence
H2O Lightning cards · JevBench 231 public items	83.1%	Setup & source Correct answers on the 231 public JevBench items, read through JevBench's own CLI in the setup each card names for JevBench. Release v1.1, stock vLLM 0.30.0, one NVIDIA L40S; serve.sh as shipped, JevBench commit bb05a33. 192 of 231 correct on the card. H2O.ai's own runs, not a leaderboard result: its Decision Index rows use the public 0.3 kit and its own scorer, and the JevBench rows use JevBench's own CLI on the 231 public items. Hardware and serving profile differ by release, as each result states. Never compare with Leanmcp JevBench's board rows, Benchmark Heaven or community Decision Index 0.4 rows. Not ranked here. Checked / reported: 2026-11-02 · Samples: 231 Snapshot: 2026-11-02 · Revision: Cards read 2026-11-02; each result names its release and card revision · Published observations; result reuse terms not specified Result source · Benchmark source
H2O Lightning cards · Decision Index 0.3 public text index	53.20	Setup & source The kit's chance-corrected index on the public text suite (140,178 scored rows), in the setup each card names for the Decision Index. Release v1.1, stock vLLM 0.30.0, one NVIDIA L40S; serve.sh as shipped: 40,960-token context, 12,000-token state budget. Kit at commit f8552a1. Printed on the model card. H2O.ai's own runs, not a leaderboard result: its Decision Index rows use the public 0.3 kit and its own scorer, and the JevBench rows use JevBench's own CLI on the 231 public items. Hardware and serving profile differ by release, as each result states. Never compare with Leanmcp JevBench's board rows, Benchmark Heaven or community Decision Index 0.4 rows. Not ranked here. Checked / reported: 2026-11-02 · Samples: 140178 Snapshot: 2026-11-02 · Revision: Cards read 2026-11-02; each result names its release and card revision · Published observations; result reuse terms not specified Result source · Benchmark source
H2O Lightning cards · Nemotron Nano, no state trimming · Decision Index 0.3 public text index	53.22	Setup & source H2O-Lightning-4B-Nemotron-Nano with a 131,072-token context and no state trimming. Release v1.1, stock vLLM 0.30.0, one NVIDIA L40S; no state trimming: MAX_LEN=131072, SHIM_MAX_STATE_TOKENS=1000000. Kit at commit f8552a1. Printed on the model card. H2O.ai's own runs, not a leaderboard result: its Decision Index rows use the public 0.3 kit and its own scorer, and the JevBench rows use JevBench's own CLI on the 231 public items. Hardware and serving profile differ by release, as each result states. Never compare with Leanmcp JevBench's board rows, Benchmark Heaven or community Decision Index 0.4 rows. Not ranked here. Checked / reported: 2026-11-02 · Samples: 140178 Snapshot: 2026-11-02 · Revision: Cards read 2026-11-02; each result names its release and card revision · Published observations; result reuse terms not specified Result source · Benchmark source