Can I run it?
Can I run Qwen3 Coder 30B A3B Instruct on Apple M5 Pro?
Tight fit
Tight fit fully in unified memory on your Apple M5 Pro at UD-Q4_K_XL, 4K context, about 22–30 tok/s. Weights 16.5 GB plus 0.38 GB of KV cache — 17.3 GB of 18.0 GB usable.
Best quantUD-Q4_K_XL
Working context4K
Max that fits4K
Speed22–30 tok/sEstimate from memory bandwidth, not a benchmark. Method.
Weights16.5 GB
KV cache0.38 GB
Total17.3 GB
Usable18.0 GB
Qwen3 Coder 30B A3B Instruct details · Apple M5 Pro page · How this estimate is made
Run commands
llama.cpp
llama-server -hf unsloth/Qwen3-Coder-30B-A3B-Instruct-GGUF:UD-Q4_K_XL -c 4096 -ngl 48
Ollama
ollama run qwen3
LM Studio
lms get unsloth/Qwen3-Coder-30B-A3B-Instruct-GGUF # Then load quant UD-Q4_K_XL
Assumptions
- Estimate from memory bandwidth, not a benchmark. Decode speed uses 55–75% of peak bandwidth.
- KV cache is FP16. Engine overhead assumes llama.cpp (0.3 GB + 1% of the weights). Ollama is closer to 0.9 GB + 3%, LM Studio closer to 1 GB + 3%.
- UD-Q4_K_XL size is the real GGUF file size from Hugging Face.
- Unified memory uses 75% for the GPU by default, and always leaves about 5 GB for the OS.
- Speed is quoted at 4K context, not at the maximum context that still fits.
Larger models that also fit
- Can I run Cyber Tiel Coder 35B A3B GGUF MTP on Apple M5 Pro? — Tight fit, Q3_K_M, 23–32 tok/s (estimate)
- Can I run Kwaipilot KAT Coder V2.5 Dev on Apple M5 Pro? — Tight fit, Q3_K_L, 24–33 tok/s (estimate)
- Qwen2.5 Coder 32B Instruct — Tight fit, Q3_K_L, 9.2–13 tok/s (estimate)
Updated 2026-10-24. Profile on this page is the device default.