I RL-trained Qwen3.6-35B-A3B to RL-train small task-specific Qwen models. Fully open source! 🤓
Summary
The author trained a Qwen3.6-35B-A3B model using reinforcement learning to then RL-train small task-specific Qwen models, and has released everything fully open source.
Similar Articles
Training frontier knowledge work agents: A 397B RL training guide with SkyRL (18 minute read)
Mercor details the reinforcement learning post-training of Qwen3.5-397B-A17B and Qwen3.6-35B-A3B for knowledge work agents, achieving a 70% relative improvement on the APEX-Agents benchmark, and open-sources the full training recipe.
Qwen 3.8 27B is out: open weights, best local dense model yet
Qwen releases Qwen3.8-27B, an open-weights 27B dense vision-language model with major gains in coding, professional work, and long-horizon agentic tasks, available in FP8 with flexible thinking control.
Qwen 3.7 Max
Qwen 3.7 is an impressive new AI model from Chinese labs, with discussion on whether weights will be available for download.
@sgl_project: The king of small models is back! Qwen3.8-27B from @Alibaba_Qwen is open source, and Day-0 support is live in SGLang: -…
Qwen3.8-27B, a 27B-parameter multimodal AI model from Alibaba, is now open source with day-0 support in SGLang, offering high inference speeds and superior performance in coding and office tasks.
Qwen is never going to open source Qwen 3.7, aren't they?
After firing Junyang Lin, Qwen has locked down its large models and is no longer releasing open source models, while other Chinese AI labs continue to open source their latest models. Rumors suggest the small model team is gone and Qwen 3.6/3.7 may be the last open source models.