How we got this number
Each generated token reads the whole model from memory once: 49 GB. The Mac mini M5 Pro (24GB) moves 307 GB/s, so the ceiling is 6.3 tokens/sec. Real machines reach a fraction of that ceiling; we use 80%, calibrated against a measured Mac run. Send us your real numbers and we'll replace the estimate.
Faster machines for Qwen2.5-VL 72B
Other models on the Mac mini M5 Pro (24GB)
| gpt-oss 20B | Fast | 60 tok/s |
| Qwen 3 30B (A3B) | Fast | 43 tok/s |
| Qwen3 Coder 30B (A3B) | Fast | 43 tok/s |
| Gemma 4 26B (A4B) | Fast | 29 tok/s |
| starCoder 2 15B | Fast | 25 tok/s |
| Mistral small 22B | Comfortable | 19 tok/s |