How we got this number
Each generated token reads the active part of the model from memory once: 2.8 GB. The Beelink SER7 (Ryzen 7 7840HS, 32GB) moves 89.6 GB/s, so the ceiling is 32 tokens/sec. Real machines reach a fraction of that ceiling; we use 60%, an assumption until we get measured runs for this kind of machine. Send us your real numbers and we'll replace the estimate.
Faster machines for gpt-oss 120B
| Mac Studio M5 Ultra (96GB) | 120+ tok/s | Check price → |
| Mac Studio M5 Ultra (256GB) | 120+ tok/s | Check price → |
| NVIDIA DGX Spark (128GB) | 33 tok/s | Check price → |
| Ryzen AI Max+ 395 mini PC (128GB) | 31 tok/s | Check price → |
Other models on the Beelink SER7 (Ryzen 7 7840HS, 32GB)
| Qwen 3 30B (A3B) | Comfortable | 16 tok/s |
| Qwen3 Coder 30B (A3B) | Comfortable | 16 tok/s |
| Qwen 3-VL 30B (A3B) | Comfortable | 15 tok/s |
| Qwen3.5 35B (A3B) | Comfortable | 14 tok/s |
| Qwen3.6 35B (A3B) | Comfortable | 14 tok/s |
| gpt-oss 20B | Comfortable | 13 tok/s |