Qwen: Qwen3.6 Max Preview

Qwen3.6 Max Preview is a text-in, text-out model from Qwen with a 262,144-token context window and a maximum of 65,536 output tokens per response. It supports tool use and reasoning, making it suitable for multi-step tasks and agentic workflows. Structured output support is unconfirmed, so developers who depend on guaranteed JSON schema compliance should verify that independently before committing. At $1.04 per million input tokens and $6.24 per million output tokens, the output cost sits in the mid-to-upper tier of available models. Its blended benchmark score of 66.2 is drawn from a single benchmark, so competitive standing against other models is not yet well established. Buyers who specifically need long-context reasoning with tool support and are willing to accept limited benchmark evidence may find it worth testing, but those prioritizing proven benchmark coverage or lower output costs have better-documented alternatives to consider.