Can I run it?
Can I run gpt oss 20b on Apple M4 Max (40-core GPU)?
Runs great
Runs great fully in unified memory on your Apple M4 Max (40-core GPU) at UD-Q8_K_XL, 8K context, about 74–102 tok/s. Weights 12.3 GB plus 0.19 GB of KV cache — 12.9 GB of 36.0 GB usable. Up to 128K context still fits.
Best quantUD-Q8_K_XL
Working context8K
Max that fits128K
Speed74–102 tok/s
Weights12.3 GB
KV cache0.19 GB
Total12.9 GB
Usable36.0 GB
gpt oss 20b details · Apple M4 Max (40-core GPU) page · How this estimate is made
Run commands
llama.cpp
llama-server -hf unsloth/gpt-oss-20b-GGUF:UD-Q8_K_XL -c 8192 -ngl 24
No matching name in the Ollama library snapshot.
LM Studio
lms get unsloth/gpt-oss-20b-GGUF # Then load quant UD-Q8_K_XL
Assumptions
- Estimate, not a benchmark. Decode speed uses a memory-bandwidth model at 55–75% of peak.
- 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-Q8_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 8K context, not at the maximum context that still fits.
Larger models that also fit
- Can I run Qwen3.8 27B on Apple M4 Max (40-core GPU)? — Runs, UD-Q8_K_L, 11–14 tok/s
- Can I run Qwen3.6 35B A3B on Apple M4 Max (40-core GPU)? — Tight fit, Q8_0, 89–122 tok/s
- Can I run Qwen3.8 35B A3B Distill on Apple M4 Max (40-core GPU)? — Runs, Q6_K, 114–155 tok/s
Updated 2026-10-24. Profile on this page is the device default.