Apple M3 Ultra (80-core GPU, 512GB) × Mistral Small 3.1 24B
Can the Apple M3 Ultra (80-core GPU, 512GB) run Mistral Small 3.1 24B?
Yes
Yes. The Apple M3 Ultra (80-core GPU, 512GB) can run Mistral Small 3.1 24B at up to F16 (needs ~51.6 GB of 384 GB usable), at an estimated ~24.2 tokens/sec at Q4_K_M.
Speed is estimated from memory bandwidth calibrated with a measured 92.14 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 |
|---|---|---|---|
| Q3_K_M | 14.2 GB | ✅ | ~30.1 |
| Q4_K_M | 17.1 GB | ✅ | ~24.2 |
| Q5_K_M | 19.8 GB | ✅ | ~20.6 |
| Q6_K | 22.5 GB | ✅ | ~17.8 |
| Q8_0 | 28.4 GB | ✅ | ~13.8 |
| F16 | 51.6 GB | ✅ | ~7.3 |
Fastest hardware for Mistral Small 3.1 24B
- NVIDIA GeForce RTX 5090 — ~76.2 tok/s at Q4_K_M
- NVIDIA GeForce RTX 5080 — ~59.5 tok/s at Q3_K_M
- NVIDIA GeForce RTX 5070 Ti — ~57.8 tok/s at Q3_K_M
- NVIDIA GeForce RTX 4090 — ~48.9 tok/s at Q4_K_M
- NVIDIA GeForce RTX 4080 SUPER — ~48.5 tok/s at Q3_K_M
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.