MODEL / ALIBABA QWEN

Qwen3.6 35B-A3B

All 35B parameters must be stored even though approximately 3B are active per token.

一次情報

総パラメータ35B3B active / token

推定メモリ~25 GBQ4 / NVFP4 MoE

コンテキスト262K nativeText + Image

LicenseApache-2.0Multilingual

01 / 用途

fast agents
coding
tool use

02 / 推定適合

Mac Studio M5 Ultra · 96 GB余裕あり MacBook Pro M5 Max · 64 GB余裕あり Framework Desktop · Ryzen AI Max+ 395 · 128 GB余裕あり Radeon AI PRO R9700 · 32 GBぎりぎり NVIDIA DGX Spark · 128 GB余裕あり

03 / 実行環境

These are runtime examples, not validated installation recipes. Fit estimates assume quantized files; official repositories may hold full-precision weights. MLX, GGUF and TensorRT require matching conversions or engines.

vLLMAdvancedvllm serve Qwen/Qwen3.6-35B-A3B --dtype auto

llama.cppMediumllama-server -m /path/to/verified-model.gguf --port 9931

LM StudioEasySearch “Qwen/Qwen3.6-35B-A3B” → Load a Q4/INT4 build → Start Server

OllamaEasyollama run <model-tag>

注意

All 35B parameters must be stored even though approximately 3B are active per token.