How we got this number
Each generated token reads the active part of the model from memory once: 2.4 GB. The NVIDIA DGX Spark (128GB) moves 273 GB/s, so the ceiling is 114 tokens/sec. Real machines reach a fraction of that ceiling; we use 70%, 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 20B
Other models on the NVIDIA DGX Spark (128GB)
| lfm 2.5 8B (A1B) | Fast | 120+ tok/s |
| Qwen2.5-VL 3B | Fast | 60 tok/s |
| Gemma 3 4B | Fast | 58 tok/s |
| Qwen 3-VL 4B | Fast | 58 tok/s |
| translateGemma 4B | Fast | 58 tok/s |
| Qwen3.5 4B | Fast | 56 tok/s |