LocalForge
Back to guide
MODEL / ALIBABA QWEN

Qwen3.6 35B-A3B

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

Primary source
Total parameters35B3B active / token
Estimated memory~25 GBQ4 / NVFP4 MoE
Context262K nativeText + Image
LicenseQwen CommunityMultilingual
01 / BEST FOR
  • fast agents
  • coding
  • tool use
02 / ESTIMATED FIT
Mac Studio M5 Ultra · 96 GBComfortableMacBook Pro M5 Max · 64 GBComfortableFramework Desktop · Ryzen AI Max+ 395 · 128 GBComfortableRadeon AI PRO R9700 · 32 GBTightNVIDIA DGX Spark · 128 GBComfortable
03 / DEPLOY WITH
vLLMAdvancedvllm serve Qwen/Qwen3.6-35B-A3B --dtype auto
llama.cppMediumllama-server -m /path/to/verified-model.gguf --port 8080
LM StudioEasySearch “Qwen/Qwen3.6-35B-A3B” → Load a Q4/INT4 build → Start Server
OllamaEasyollama run <model-tag>
Watch out

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