LLM tokens per second — live multi-provider benchmark
Real inference speed, measured continuously. Every row is a live model from Ollama, OpenCode Zen, or OpenCode Go — sorted by tokens per second, benchmarked every ~10 minutes on Ollama Pro, every ~60 minutes on Ollama Free and OpenCode Zen, and every ~30 minutes on OpenCode Go for its standard-tier models.
Independent & vendor-neutral — no paid placement. TokenDyno is not affiliated with any provider it benchmarks and receives no payment for ranking or favorable coverage. Read the independence policy →
| Trend | ||||||||
|---|---|---|---|---|---|---|---|---|
| Claude Haiku 5 5 | OpenCode Go | 536.9 | 528.7 | 2.9s | 75% n=4 | 43.4 | | 3h ago |
| Beam (Free) | OpenCode Go | 442.6 | 468.9 | 591ms | 100% n=4 | — | | 5h ago |
| Beam | Ollama Pro | 379.9 | 352.7 | 409ms | 99% | — | | 2m ago |
| Beam | Ollama Free | 343.7 | 308.1 | 533ms | 91% n=11 | — | | 1m ago |
| GPT-OSS 120B | Ollama Free | 342.1 | 346.0 | 333ms | 100% | 11.6 | | 27m ago |
| GPT-OSS 120B | Ollama Pro | 262.3 | 294.4 | 332ms | 100% | 11.6 | | 5m ago |
| Muse Spark Open | Ollama Pro | 243.7 | 131.5 | — | 100% | 38.9 | | 6m ago |
| Gemma4 31B | Ollama Pro | 242.3 | 236.6 | 338ms | 100% | 14.7 | | 2m ago |
| Nemotron 3 Nano 30B (non-reasoning) | Ollama Free | 222.1 | 284.7 | 318ms | 100% | 6.8 | | 24m ago |
| Nemotron 3 Nano 30B (non-reasoning) | Ollama Pro | 212.4 | 331.2 | 291ms | 99% | 6.8 | | 2m ago |
| Gemma4 31B | Ollama Free | 209.9 | 43.4 | 365ms | 100% | 14.7 | | 29m ago |
| Muse Spark 1.3 Contributor | OpenCode Go | 197.8 | 113.9 | 1.6s | 75% n=4 | 48.1 | | 1h ago |
| Space Bunny | OpenCode Go | 194.1 | 175.0 | 974ms | 100% n=4 | — | | 3h ago |
| DeepSeek V4.1 Flash | Ollama Pro | 186.7 | 125.1 | 2.3s | 100% | 39.5 | | 5m ago |
| Kimi K2.6 | Ollama Pro | 174.0 | 116.5 | 605ms | 100% | 27 | | 7m ago |
| GLM 5.3 Flash | Ollama Pro | 167.1 | 145.4 | 825ms | 100% | 41.8 | | 6m ago |
| MiniMax M3 | Ollama Pro | 163.1 | 132.3 | 1.5s | 100% | 29.2 | | 4m ago |
| DeepSeek V4.1 Flash | OpenCode Go | 161.9 | 180.1 | 1.4s | 100% n=4 | 39.5 | | 1h ago |
| Step 5 Preview | OpenCode Go | 155.1 | 161.4 | 1.4s | 100% n=4 | 43.7 | | 15m ago |
| DeepSeek V4 Pro 0813 | Ollama Pro | 141.2 | 196.7 | 657ms | 100% | 36 | | 8m ago |
| MiniMax M3 | OpenCode Go | 136.3 | 99.3 | 1.2s | 100% n=4 | 29.2 | | 2h ago |
| GPT-OSS 20B | Ollama Pro | 133.7 | 109.0 | 499ms | 100% | 9 | | 8m ago |
| GPT-OSS 20B | Ollama Free | 131.1 | 118.8 | 434ms | 100% | 9 | | 25m ago |
| Kimi K2.7 Code | Ollama Pro | 129.6 | 133.1 | 1.5s | 100% | 25.8 | | 7m ago |
| GLM 5.3 | Ollama Pro | 123.1 | 181.8 | 720ms | 100% | 44.8 | | 5m ago |
| GPT 5.6 Luna | OpenCode Go | 116.3 | 117.6 | 980ms | 75% n=4 | 37.3 | | 1h ago |
| GPT 6 Luna | OpenCode Go | 114.7 | 102.1 | 1.1s | 50% n=4 | 38.1 | | 1h ago |
| Qwen3.7 Max | OpenCode Go | 111.3 | 104.6 | 1.5s | 100% n=4 | 29.5 | | 3h ago |
| Grok 4.6 | OpenCode Go | 92.0 | 92.4 | 71.2s | 75% n=4 | 44.3 | | 4h ago |
| Kimi K3 | Ollama Pro | 91.6 | 97.8 | 1.1s | 100% n=6 | 43.6 | | 1h ago |
| DeepSeek V4 Pro | OpenCode Go | 87.3 | 78.6 | 1.4s | 100% n=4 | 36 | | 3h ago |
| GLM 5.3 | OpenCode Go | 80.1 | 99.5 | 802ms | 100% n=4 | 44.8 | | 4h ago |
| Hy3 | OpenCode Go | 78.8 | 88.1 | 1.7s | 100% n=4 | 25.3 | | 3h ago |
| GLM 5.2 | Ollama Pro | 76.1 | 63.6 | 650ms | 100% | 33.7 | | 3m ago |
| Mistral Large 3 675B (non-reasoning) | Ollama Pro | 73.2 | 69.5 | 669ms | 99% | 9.3 | | 6m ago |
| GLM 5.2 | OpenCode Go | 70.8 | 89.0 | 2.0s | 100% n=4 | 33.7 | | 3h ago |
| Nemotron 3 Ultra | Ollama Free | 66.8 | 39.8 | 2.9s | 100% | 22.9 | | 17m ago |
| GLM 5.1 | OpenCode Go | 66.8 | 76.5 | 877ms | 100% n=4 | 26.1 | | 3h ago |
| MiniMax M2.5 | OpenCode Go | 65.5 | 63.7 | 890ms | 100% n=4 | 22.8 | | 2h ago |
| MiniMax M2.7 | Ollama Pro | 64.8 | 52.8 | 1.2s | 100% | 22.8 | | 5m ago |
| Muse Spark 1.2 Contributor | OpenCode Go | 63.9 | 79.4 | 2.8s | 100% n=4 | 39.6 | | 42m ago |
| MiniMax M2.7 | OpenCode Go | 62.9 | 70.7 | 884ms | 100% n=4 | 22.8 | | 3h ago |
| Qwen3.7 Plus | OpenCode Go | 62.5 | 62.5 | 1.5s | 75% n=4 | 25.2 | | 2h ago |
| Grok 4.7 | OpenCode Go | 61.7 | 67.2 | 4.4s | 100% n=4 | 46.4 | | 3h ago |
| Mistral Large 4 | Ollama Pro | 60.9 | 81.9 | 763ms | 100% | 38.4 | | 10m ago |
| Nemotron 3 Ultra | Ollama Pro | 60.3 | 61.0 | 4.0s | 100% | 22.9 | | 7m ago |
| Qwen3.6 Plus | OpenCode Go | 58.8 | 58.8 | 1.0s | 75% n=4 | 27 | | 3h ago |
| Longcat 2.0 | OpenCode Go | 58.8 | 61.1 | 80.4s | 100% n=4 | 19.1 | | 3h ago |
| Omen Alpha | OpenCode Go | 58.8 | 77.3 | 2.4s | 100% n=4 | — | | 2h ago |
| GLM 5.3 Flash | OpenCode Go | 52.7 | 23.2 | 1.2s | 100% n=4 | 41.8 | | 3h ago |
| Qwen3.8 Flash | OpenCode Go | 51.9 | 53.3 | 1.0s | 100% n=4 | 39.8 | | 3h ago |
| Kimi K2.7 Code | OpenCode Go | 44.6 | 33.3 | 1.4s | 100% n=4 | 25.8 | | 3h ago |
| Kimi K2.6 | OpenCode Go | 43.9 | 46.2 | 1.3s | 100% n=4 | 27 | | 40m ago |
| Kimi K3 | OpenCode Go | 42.2 | 43.5 | 1.9s | 100% n=4 | 43.6 | | 5h ago |
| Nemotron 3 Super | Ollama Free | 42.1 | 30.1 | 543ms | 100% | 12.8 | | 22m ago |
| Nemotron 3 Super | Ollama Pro | 41.8 | 38.0 | 642ms | 99% | 12.8 | | 4m ago |
| MiMo V2.6 Pro | OpenCode Go | 41.0 | 44.0 | 2.3s | 100% n=4 | 46.3 | | 44m ago |
| Longcat 2.5 Preview (Free) | OpenCode Go | 40.3 | 34.9 | 1.8s | 100% n=4 | — | | 3h ago |
| MiMo V2.5 Pro | OpenCode Go | 39.7 | 37.3 | 3.2s | 100% n=4 | 26 | | 3h ago |
| Qwen3.8 Max | OpenCode Go | 35.6 | 33.1 | 1.4s | 100% n=4 | 45.4 | | 2h ago |
| MiMo V2.5 | OpenCode Go | 35.4 | 33.4 | 2.4s | 100% n=4 | 21.8 | | 3h ago |
| Hy4 Preview | OpenCode Go | 31.5 | 33.3 | 1.8s | 100% n=4 | — | | 53m ago |
| MiMo V2.6 Flash | OpenCode Go | 23.6 | 23.6 | 8.1s | 100% n=1 | 37.9 | | 4h ago |
| Longcat 2.5 Preview (Free) | OpenCode Zen | — | — | — | 0% n=1 | — | | — |
| Beam (Free) | OpenCode Zen | — | — | — | 0% | — | | — |
| Space Bunny (Free) | OpenCode Zen | — | 172.0 | 758ms | 0% n=1 | — | | 6d ago |
| Nemotron 3.5 Lightning (Free) | OpenCode Zen | — | 57.0 | 651ms | 0% n=1 | 12.9 | | 30d ago |
| Nemotron 3.5 Super Vl (Free) | OpenCode Zen | — | — | — | 0% | 26.5 | | — |
| MiMo V2.6 Flash (Free) | OpenCode Zen | — | — | — | 0% n=1 | 37.9 | | — |
| Big Pickle | OpenCode Zen | — | 50.3 | 697ms | 0% n=1 | — | | 32d ago |
| DeepSeek V4.1 Flash (Free) | OpenCode Zen | — | — | — | 0% | 39.5 | | — |
| GLM 5.3 Flash (Free) | OpenCode Zen | — | — | — | 0% | 41.8 | | — |
| Muse Spark 1.3 Contributor (Free) | OpenCode Zen | — | 54.4 | 6.3s | 0% n=1 | 48.1 | | 30d ago |
No models match your filter.
Intelligence Index scores from Artificial Analysis.
Ollama Free and OpenCode Zen are sampled about hourly to avoid burning through their plan balances. OpenCode Go is sampled about every 30 minutes for its standard-tier models, while its most expensive models — and any Go model not yet cleared for the faster tier — are held at about every 6 hours; sampling may pause or slow below those configured cadences while the shared Go plan budget is conserved. Ollama Pro is sampled about every 10 minutes.
10 models unavailable or stale
Support the live benchmarks
Frequently asked questions
What is the best LLM tokens per second leaderboard?
There is no single "best" leaderboard because different tools measure different things. TokenDyno is a live tokens-per-second tracker: it re-benchmarks Ollama Pro, Ollama Free, OpenCode Zen, and OpenCode Go models on a fixed schedule — about every 10 minutes on Ollama Pro, about every 60 minutes on Ollama Free and OpenCode Zen, and about every 30 minutes on OpenCode Go for its standard-tier models — using the same prompt and measurement method across providers, so speed numbers are directly comparable. For a capability-only score, see LiveBench; for a combined capability-speed-price view, see Artificial Analysis. See the /vs page for a factual side-by-side.
What is TokenDyno's benchmark methodology?
TokenDyno sends a fixed streaming chat-completion request (max_tokens capped at 300) to each provider and measures generation throughput using a hybrid method: inter-token timing when a response streams smoothly (≥8 chunks, mean gap >50ms), or a wall-clock fallback for burst/chunked delivery. Time-to-first-token, error taxonomy, and reliability are also recorded. Full detail is on the /methodology page.
How often is TokenDyno's data refreshed?
The worker benchmarks continuously using a round-robin priority queue. Cadence follows how each provider bills: our Ollama Pro key is a legacy flat-rate subscription, so it is sampled about every 10 minutes. Ollama Free and OpenCode Zen run on capped plans where every sample spends budget, so they are sampled about every 60 minutes. OpenCode Go is sampled about every 30 minutes for its standard-tier models, while its most expensive models — and any Go model not yet cleared for the faster tier — are held at about every 6 hours. Go sampling may pause or slow below those configured cadences while the shared Go plan budget is being conserved (the meter reflects account-level usage, which may include other usage beyond this benchmark). The leaderboard also polls the API roughly every 60 seconds client-side to keep cells current without a full page reload.