AnalysisAI ModelsSeptember 29, 2026

Open-weight 0.8B/2B decision models hit ~30 ms per decision

Read original source →reddit.com

Qwen3.5 and Gemma fine-tunes for zero-shot classification, released as open weights, match Jev on benchmarks and Doom at roughly 30 ms per decision. Give them a situation and a list of options.

2 sources

More stories today

Open the live feed