Qwen launches top-end proprietary 3.6-Max-Preview model

While Qwen is known for open weights, keeping their new flagship proprietary signals a push to compete directly with frontier models like Claude 3.5 Sonnet and GPT-4o. The model focuses heavily on agentic coding and complex tool use, key areas for developer adoption. It demonstrates large benchmark gains over the previous Qwen3.6-Plus iteration. Remaining in preview and closed-weight diverges from their popular open-source strategy for lower-tier models.