Rent a GPU

Can the Apple M3 Max (64GB) run gpt-oss 120B?

Apple M3 Max (64GB) × gpt-oss 120B8K context
Yes, with heavy quantizationFits only at a low-quality quant
Memory at Q2_K45 GB of 48 GB
Weights 43.8 GBKV cache 0.6 GBRuntime 0.6 GB
QuantQ2_KHeavily degraded
Memory45 GBof 48 GB usable
Speed~125tokens per second
  1. FP16
  2. Q8_0
  3. Q6_K
  4. Q5_K_M
  5. Q4_K_M
  6. Q3_K_M
  7. Q2_K

Only just. gpt-oss 120B fits on the Apple M3 Max (64GB) at Q2_K, which costs noticeable quality. Expect about 125 tokens/second. A card with more memory is a better experience.

Try other context lengths

Run it with Ollama

Ollama's default gpt-oss:120b download is Q4_K_M, which needs about 72 GB, more than the 48 GB available. Download the Q2_K GGUF instead (LM Studio lists quants per model), or try a smaller context length.

Run it in the cloud

Want full quality? Run gpt-oss 120B on a bigger rented GPU.

Rent a cloud GPU by the hour. No upfront cost, stop any time.

Rent a GPU on RunPodRent a GPU on Vast.ai

Memory needed by quantization

At 8,192 tokens of context. The black line marks the 48 GB the Apple M3 Max (64GB) can use. macOS lets the GPU use about 48 GB of the 64 GB unified memory by default.

QuantQualityMemoryFits?Speed
FP16Full quality234.8 GBToo big—
Q8_0Near-lossless125.3 GBToo big—
Q6_KExcellent97 GBToo big—
Q5_K_MVery good84.3 GBToo big—
Q4_K_MGood, most popular72 GBToo big—
Q3_K_MNoticeably degraded58.3 GBToo big—
Q2_KHeavily degraded45 GBFits~125 tok/s

GPUs that run gpt-oss 120B comfortably

Models that run well on the Apple M3 Max (64GB)

Frequently asked questions

Can Apple M3 Max (64GB) run gpt-oss 120B?

Only just. gpt-oss 120B fits on the Apple M3 Max (64GB) at Q2_K, which costs noticeable quality. Expect about 125 tokens/second. A card with more memory is a better experience.

How much VRAM does gpt-oss 120B need?

About 72 GB at Q4_K_M with 8,192 tokens of context, and about 234.8 GB at full FP16 precision.

How fast is gpt-oss 120B on the Apple M3 Max (64GB)?

Roughly 125 tokens per second at Q2_K, estimated from the Apple M3 Max (64GB)'s 400 GB/s memory bandwidth.