H2O-Lightning-4B is faster and more capable than both OpenAI Decisions and Microsoft Decisions. huggingface.co/h2oai/h2o-lig…
1
3
A real external JevBench measurement of Microsoft-Decision-1 latency does not agree with Microsoft's numbers. Also independent accuracy does not agree with their rankings as well. See
Corrected latency for the Microsoft animation. Also, accuracy presented is not an independent benchmark like JevBench, so training and test may be too similar since generated by Microsoft themselves.
1
5
1,990
Model (Apache-2.0): huggingface.co/h2oai/h2o-lig…
Blog: h2o.ai/blog/2026/h2o-lightni…
Full demos: youtu.be/2Qp04Wu0A14
Leaderboard: benchmarkheaven.com/jev-mode…
1
1
47
🚨 H2O-Lightning-4B is #1 among open-weight models on JevBench, ahead of Jev itself.
72.5 vs 71.5 on the official composite (intelligence, calibration, cost and speed), ahead of every open 12B, 26B and 31B model.
4B parameters. Apache-2.0. 🧵
2
3
210
1 year later, LLMs get 65% on test set vs. degree-holding humans at 92%. Not bad year I think.
h2o.ai/blog/2024/h2o-ai-tops…
18
