Apple M3 Ultra (80-core GPU, 512GB) × DeepSeek V3 / R1 671B (MoE)
Can the Apple M3 Ultra (80-core GPU, 512GB) run DeepSeek V3 / R1 671B (MoE)?
Yes, usable
Yes. The Apple M3 Ultra (80-core GPU, 512GB) can run DeepSeek V3 / R1 671B (MoE) at up to Q3_K_M (needs ~339 GB of 384 GB usable), at an estimated ~9.3 tokens/sec at Q3_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 | 339 GB | ✅ | ~9.3 |
| Q4_K_M | 421 GB | ❌ | — |
| Q5_K_M | 494 GB | ❌ | — |
| Q6_K | 572 GB | ❌ | — |
| Q8_0 | 736 GB | ❌ | — |
| F16 | 1384 GB | ❌ | — |
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.