Qwen3.6-27B tops open-weight intelligence leaderboard

This is the kind of result that matters more than a giant-parameter vanity win: a dense 27B open-weight model is now credible enough to sit at the front of the pack. The practical story is deployment leverage, not just benchmark bragging rights.

• –Artificial Analysis puts Qwen3.6 27B (Reasoning) at #1 among 124 models on the Intelligence Index with a score of 46.
• –Hugging Face says the release improves agentic coding and adds "thinking preservation," which should help iterative development workflows.
• –The model ships with 262K native context and support for text, image, and video input, making it more than a narrow code model.
• –Apache 2.0 weights plus support for vLLM, SGLang, and other open tooling make it easy to slot into self-hosted stacks.
• –The tradeoff is cost and speed: Artificial Analysis flags it as relatively slow, expensive, and verbose versus other open-weight options.