Apple M1 Max (32-core GPU, 64GB) × gpt-oss-20b (MoE)
Can the Apple M1 Max (32-core GPU, 64GB) run gpt-oss-20b (MoE)?
Yes
Yes. The Apple M1 Max (32-core GPU, 64GB) can run gpt-oss-20b (MoE) at up to MXFP4 (native) (needs ~14.6 GB of 48.0 GB usable), at an estimated ~56.2 tokens/sec at MXFP4 (native).
Speed is estimated from memory bandwidth calibrated with a measured 61.19 tok/s Llama 2 7B Q4_0 result (source). No direct public benchmark of this exact pair is recorded yet.
By quantization (8K context)
| Quant | Memory needed | Fits? | Est. tok/s |
|---|---|---|---|
| MXFP4 (native) | 14.6 GB | ✅ | ~56.2 |
Fastest hardware for gpt-oss-20b (MoE)
- NVIDIA GeForce RTX 5090 — ~267 tok/s at MXFP4 (native)
- NVIDIA GeForce RTX 4090 — ~171 tok/s at MXFP4 (native)
- AMD Radeon RX 7900 XTX — ~169 tok/s at MXFP4 (native)
- NVIDIA GeForce RTX 5080 — ~167 tok/s at MXFP4 (native)
- NVIDIA GeForce RTX 5070 Ti — ~163 tok/s at MXFP4 (native)
No suitable GPU? Rent one by the hour on Vast.ai or RunPod (affiliate links).
Get new models & benchmarks in your inbox
No spam. Unsubscribe anytime by replying to any email.