Genspark and Baseten Post-Train Qwen3.5-397B to Beat Opus 5.5 in Deep Research

Genspark and Baseten jointly post-trained the open-weights Qwen3.5-397B, claiming it beats Claude Opus 5.5 on internal Deep Research benchmarks. Their method splits the task into RL-trained search and SFT-based report writing.

2026-11-11 ~ 2026-11-11 · 2 related posts

• Genspark and Baseten post-train open-weight Qwen3.5-397B to beat Opus 5.5 at deep research — baseten · 2026-11-11
• How Genspark's Opus-beating model was trained: breadth-rewarding RL for research, SFT plus RL for reports — baseten · 2026-11-11