Prime Intellect Open-Sources Full Post-Training Stack as Extropic Nearly Triples Qwen3.6 Benchmarks

Prime Intellect has opened its full post-training stack, partnering with Extropic to custom-train models. Using about 100 RL steps, Extropic nearly tripled Qwen3.6-35B-A3B's benchmark scores for thermodynamic ML research.

2026-10-02 ~ 2026-10-02 · 2 related posts

• Prime Intellect ships full post-training stack as Extropic runs custom RL with it — beffjezos · 2026-10-02
• Extropic post-trains Qwen3.6-35B-A3B on Prime Intellect, nearly tripling evals in ~100 GRPO steps — MarvinTBaumann · 2026-10-02