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 | 311.7 | 432.1 | 342ms | 100% | 11.6 | | 26m ago |
| GPT-OSS 120B Pro | 280.2 | 229.6 | 430ms | 100% | 11.6 | | 2m ago |
| Nemotron 3 Nano 30B (non-reasoning) Free | 230.3 | 250.0 | 320ms | 100% | 6.8 | | 26m ago |
| Gemma4 31B Pro | 218.4 | 144.1 | 338ms | 100% | 14.7 | | 1m ago |
| Gemma4 31B Free | 201.4 | 237.0 | 331ms | 100% | 14.7 | | 27m ago |
| DeepSeek V4.1 Flash Pro | 199.2 | 279.2 | 288ms | 100% | 39.5 | | 2m ago |
| DeepSeek V4.1 Flash 1017 Pro | 198.0 | 223.9 | 645ms | 100% | 45.1 | | 2m ago |
| GLM 5.2 Pro | 180.2 | 173.9 | 554ms | 100% | 33.7 | | 1m ago |
| Nemotron 3 Nano 30B (non-reasoning) Pro | 167.8 | 311.6 | 377ms | 100% | 6.8 | | 1m ago |
| Nemotron 3 Super Pro | 162.9 | 160.0 | 461ms | 100% | 12.8 | | 1m ago |
| GLM 5.3 Flash Pro | 161.9 | 115.2 | 662ms | 100% | 41.8 | | 2m ago |
| Beam Pro | 154.9 | 151.1 | 428ms | 100% | 30.3 | | 1m ago |
| Nemotron 3 Super Free | 154.6 | 135.8 | 343ms | 100% | 12.8 | | 26m ago |
| Muse Spark Open Pro | 154.1 | 38.8 | — | 99% | 38.9 | | 3m ago |
| Beam Free | 148.4 | 166.3 | 591ms | 100% | 30.3 | | 27m ago |
| DeepSeek V4 Pro 0813 Pro | 143.7 | 130.6 | 447ms | 100% | 36 | | 4m ago |
| GPT-OSS 20B Free | 123.8 | 122.6 | 368ms | 100% | 9 | | 26m ago |
| Kimi K2.7 Code Pro | 123.7 | 119.9 | 683ms | 100% | 25.8 | | 4m ago |
| GPT-OSS 20B Pro | 119.4 | 130.5 | 408ms | 100% | 9 | | 4m ago |
| MiniMax M3.1 Flash Pro | 115.0 | 135.2 | 861ms | 100% | 42.7 | | 1m ago |
| MiniMax M3 Pro | 113.7 | 143.0 | 797ms | 100% | 29.2 | | 1m ago |
| Kimi K2.6 Pro | 112.4 | 82.5 | 622ms | 100% | 27 | | 4m ago |
| GLM 5.3 Pro | 107.2 | 155.3 | 322ms | 100% | 44.8 | | 2m ago |
| Kimi K3 Pro | 97.0 | 92.9 | 856ms | 100% n=4 | 43.6 | | 1h ago |
| Mistral Large 4 Pro | 90.7 | 80.3 | 786ms | 100% | 38.4 | | 5m ago |
| Mistral Large 3 675B (non-reasoning) Pro | 70.2 | 80.5 | 1.1s | 94% | 9.3 | | 3m ago |
| Nemotron 3 Ultra Free | 46.5 | 45.1 | 783ms | 100% | 22.9 | | 26m ago |
| Nemotron 3 Ultra Pro | 41.5 | 51.0 | 617ms | 100% | 22.9 | | 4m ago |
| MiniMax M2.7 Pro | 37.4 | 42.5 | 641ms | 100% | 22.8 | | 2m 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 432.1 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.