Ollama Cloud tokens per second — live benchmark
Real inference speed, measured continuously. Every row is a live Ollama Cloud model — sorted by tokens per second, benchmarked every ~10 minutes.
Weighing Ollama Cloud against a router? Ollama Cloud vs OpenRouter — what we measure, and what we don't.
| Trend | |||||||
|---|---|---|---|---|---|---|---|
| GPT-OSS 120B Free | 333.9 | 362.6 | 338ms | 100% | 11.6 | | 42m ago |
| GPT-OSS 120B Pro | 290.9 | 290.8 | 362ms | 100% | 11.6 | | 52s ago |
| MiMo V2.6 Flash Pro | 256.7 | 213.1 | 309ms | 100% | 37.9 | | 3m ago |
| MiMo V2.6 Pro Pro | 236.6 | 263.9 | 697ms | 100% | 46.3 | | 3m ago |
| Gemma4 31B Pro | 235.3 | 291.4 | 291ms | 100% | 14.7 | | 6m ago |
| Nemotron 3 Nano 30B (non-reasoning) Free | 230.1 | 140.6 | 1.1s | 100% | 6.8 | | 39m ago |
| Nemotron 3 Nano 30B (non-reasoning) Pro | 227.3 | 167.2 | 256ms | 100% | 6.8 | | 6m ago |
| DeepSeek V4.1 Flash Pro | 221.5 | 200.1 | 459ms | 100% | 39.5 | | 1m ago |
| Gemma4 31B Free | 220.1 | 289.1 | 325ms | 100% | 14.7 | | 43m ago |
| GLM 5.3 Flash Pro | 217.8 | 266.6 | 497ms | 100% | 41.8 | | 1m ago |
| DeepSeek V4.1 Flash 1017 Pro | 211.4 | 216.7 | 584ms | 100% | 45.1 | | 8m ago |
| Muse Spark Open Pro | 195.3 | 190.4 | — | 100% | 38.9 | | 2m ago |
| GLM 5.2 Pro | 185.0 | 201.7 | 382ms | 100% | 33.7 | | 6m ago |
| Kimi K2.6 Pro | 167.8 | 189.6 | 439ms | 100% | 27 | | 3m ago |
| GLM 5.3 Pro | 161.2 | 79.3 | 441ms | 100% | 44.8 | | 8m ago |
| MiniMax M3.1 Flash Pro | 155.9 | 78.0 | 582ms | 100% | 42.7 | | 8m ago |
| DeepSeek V4 Pro 0813 Pro | 151.4 | 174.3 | 454ms | 100% | 36 | | 4m ago |
| Kimi K2.7 Code Pro | 134.0 | 128.7 | 706ms | 100% | 25.8 | | 3m ago |
| GPT-OSS 20B Free | 133.8 | 128.8 | 376ms | 100% | 9 | | 39m ago |
| Nemotron 3 Super Pro | 129.8 | 120.1 | 459ms | 100% | 12.8 | | 8m ago |
| Nemotron 3 Super Free | 126.3 | 212.9 | 378ms | 100% | 12.8 | | 37m ago |
| MiniMax M3 Pro | 120.7 | 223.6 | 755ms | 100% | 29.2 | | 8m ago |
| GPT-OSS 20B Pro | 119.8 | 135.3 | 369ms | 100% | 9 | | 4m ago |
| Kimi K3 Pro | 110.0 | 107.9 | 570ms | 100% n=6 | 43.6 | | 3h ago |
| Beam Free | 105.3 | 117.0 | 504ms | 100% | 30.3 | | 43m ago |
| Beam Pro | 100.4 | 95.3 | 421ms | 100% | 30.3 | | 6m ago |
| Mistral Large 4 Pro | 96.9 | 28.5 | 7.0s | 99% | 38.4 | | 5m ago |
| Mistral Large 3 675B (non-reasoning) Pro | 86.9 | 97.3 | 2.2s | 100% | 9.3 | | 2m ago |
| Kimi K3.1 Pro Pro | 67.7 | 67.1 | 1.9s | 100% | — | | 6m ago |
| MiniMax M2.7 Pro | 52.6 | 54.0 | 895ms | 100% | 22.8 | | 1m ago |
| Nemotron 3 Ultra Free | 47.8 | 56.9 | 1.0s | 100% | 22.9 | | 33m ago |
| Nemotron 3 Ultra Pro | 43.2 | 79.4 | 716ms | 100% | 22.9 | | 3m ago |
No models match your filter.
Intelligence Index scores from Artificial Analysis.
Ollama Free is sampled about hourly to avoid burning through the weekly free-tier balance.
0 models unavailable or stale
Support the live benchmarks
Frequently asked questions
What is the fastest Ollama Cloud model right now?
As of the last build, GPT-OSS 120B is the fastest Ollama Cloud model on this leaderboard at 362.6 tokens/sec (most recent measured run). Rankings change as models are re-benchmarked roughly every 10 minutes (every 60 minutes on the Ollama Free tier) — see the live leaderboard above for the current order.
How is tokens per second measured?
TPS is generation throughput: output tokens divided by server-reported generation time, excluding time-to-first-token. For Ollama Cloud we use the server’s own reported timing (eval_count and total_duration) rather than a client-side stopwatch, so results are immune to network jitter and token buffering. Full formula and error handling are on the methodology page.
How often is data refreshed?
The worker benchmarks continuously using a round-robin priority queue. Ollama Cloud Pro is sampled about every 10 minutes; Ollama Free is sampled about every 60 minutes to preserve the weekly free-tier quota. Kimi K3 retains a conservative target of about every 4 hours. The leaderboard also polls the API roughly every 60 seconds client-side to keep cells current without a full page reload.