MODEL / ALIBABA QWEN

Qwen3.6 27B

The memory figure is a conservative Q4 fit budget, not the official full-precision checkpoint size. Exact speed depends on the quantized build and runtime.

一次情報

総パラメータ27B

推定メモリ~18.5 GBQ4 / NVFP4 estimate

コンテキスト262K nativeText + Image

LicenseQwen CommunityMultilingual

01 / 用途

coding agents
reasoning
multimodal work

02 / 推定適合

Mac Studio M5 Ultra · 96 GBComfortable Mac mini M6 · 24 GBTight MacBook Pro M5 Max · 64 GBComfortable Framework Desktop · Ryzen AI Max+ 395 · 128 GBComfortable Radeon AI PRO R9700 · 32 GBComfortable

03 / 実行環境

vLLMAdvancedvllm serve Qwen/Qwen3.6-27B --dtype auto

llama.cppMediumllama-server -m /path/to/verified-model.gguf --port 8080

LM StudioEasySearch “Qwen/Qwen3.6-27B” → Load a Q4/INT4 build → Start Server

MLX LMMediummlx_lm.chat --model Qwen/Qwen3.6-27B

注意

The memory figure is a conservative Q4 fit budget, not the official full-precision checkpoint size. Exact speed depends on the quantized build and runtime.