Qwen3.6-27B beats Qwen3.5-397B flagship on coding benchmarks

A 27B dense model is now beating a ~397B-class flagship on coding benchmarks, and it's small enough to actually run locally. Qwen3.6-27B just landed as an open Apache 2.0 multimodal model, and the immediate shift is practical: Unsloth says GGUF builds can run on 18GB RAM.

• SWE-bench Verified: 77.2 vs 76.2 for Qwen3.5-397B-A17B
• Terminal-Bench 2.0: 59.3 vs 52.5

So a model roughly 15x smaller in total parameters is posting better coding scores than Alibaba's prior flagship, while dropping hardware needs from ~807GB class setups to something in the ~56GB range (or lower with quantization). Same general capability tier, but suddenly in reach for local runs instead of datacenter-scale setups.