Apple M3 Ultra (80-core GPU, 512GB) × Qwen3 14B
Can the Apple M3 Ultra (80-core GPU, 512GB) run Qwen3 14B?
Yes
Yes. The Apple M3 Ultra (80-core GPU, 512GB) can run Qwen3 14B at up to F16 (needs ~32.6 GB of 384 GB usable), at an estimated ~39.3 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 | 9.6 GB | ✅ | ~48.8 |
| Q4_K_M | 11.4 GB | ✅ | ~39.3 |
| Q5_K_M | 13.0 GB | ✅ | ~33.4 |
| Q6_K | 14.7 GB | ✅ | ~28.8 |
| Q8_0 | 18.3 GB | ✅ | ~22.4 |
| F16 | 32.6 GB | ✅ | ~11.9 |
Fastest hardware for Qwen3 14B
- NVIDIA GeForce RTX 5090 — ~124 tok/s at Q4_K_M
- NVIDIA GeForce RTX 4090 — ~79.3 tok/s at Q4_K_M
- AMD Radeon RX 7900 XTX — ~78.5 tok/s at Q4_K_M
- NVIDIA GeForce RTX 5080 — ~77.5 tok/s at Q4_K_M
- NVIDIA GeForce RTX 5070 Ti — ~75.3 tok/s at Q4_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.