Qwen3.6-27B: Flagship-Level Coding in a 27B Dense Model

Qwen3.6-27B claims flagship-level coding performance, surpassing Qwen3.5-397B-A17B on major coding benchmarks. The post shows a 16.8GB quantized local setup via llama-server, contrasts 55.6GB full model with 807GB of the older release, and includes sample transcripts and timing data.

Source

Simon Willison

Ingéré

22 AVR. 2026 · 19:10

Score édito

4.0 / 5