TestsBenchmarks · Long context
Artificial Analysis output speed
Median output tokens per second on the first-party API, measured by Artificial Analysis (unit is tokens/sec).
Measured by independent testersIndependent Reported by the makerVendor-reported
| Thinking levelSetting | |||||||
|---|---|---|---|---|---|---|---|
| 1 | Nemotron 3.5 Lightning 30B-A3B | NVIDIA | 298 tok/s | DefaultNemotron 3.5 Lightning | Independent testIndependentArtificial Analysis ↗ | — | 1 Oct 2026 |
| 2 | Gemini 3.7 Flash | Google (Gemini / DeepMind) | 291 tok/s | HighGemini 3.7 Flash (High) | Independent testIndependentArtificial Analysis ↗ | — | 1 Oct 2026 |
| 3 | Muse Spark 1.2 | Meta (Meta Superintelligence Labs, Muse) | 226 tok/s | Extra highMuse Spark 1.2 (Xhigh) | Independent testIndependentArtificial Analysis ↗ | — | 1 Oct 2026 |
| 4 | Gemini 3.8 Flash | Google (Gemini / DeepMind) | 221 tok/s | HighGemini 3.8 Flash (High) | Independent testIndependentArtificial Analysis ↗ | — | 1 Oct 2026 |
| 5 | DeepSeek V4 Flash Vision Exp | DeepSeek | 211 tok/s | MaxDeepSeek V4 Flash Vision (Max) | Independent testIndependentArtificial Analysis ↗ | — | 1 Oct 2026 |
| 6 | DeepSeek V4.1 Flash | DeepSeek | 209 tok/s | MaxDeepSeek V4.1 Flash (Max) | Independent testIndependentArtificial Analysis ↗ | — | 1 Oct 2026 |
| 7 | Command A+ | Cohere | 208 tok/s | DefaultCommand A+ | Independent testIndependentArtificial Analysis ↗ | — | 1 Oct 2026 |
| 8 | DeepSeek V4 Flash (0731) | DeepSeek | 204 tok/s | MaxDeepSeek V4 Flash 0731 (Max) | Independent testIndependentArtificial Analysis ↗ | — | 1 Oct 2026 |
| 9 | Mistral Small 4 | Mistral AI | 183 tok/s | HighMistral Small 4 (Reasoning) | Independent testIndependentArtificial Analysis ↗ | — | 1 Oct 2026 |
| 10 | Gemini 3.5 Flash | Google (Gemini / DeepMind) | 180 tok/s | MediumGemini 3.5 Flash (Medium) | Independent testIndependentArtificial Analysis ↗ | — | 1 Oct 2026 |
| 11 | Gemini 3.6 Flash | Google (Gemini / DeepMind) | 177 tok/s | HighGemini 3.6 Flash (High) | Independent testIndependentArtificial Analysis ↗ | — | 1 Oct 2026 |
| 12 | gpt-oss-20b | OpenAI | 174 tok/s | Highgpt-oss-20b (High) | Independent testIndependentArtificial Analysis ↗ | — | 1 Oct 2026 |
| 13 | Muse Spark 1.3 | Meta (Meta Superintelligence Labs, Muse) | 174 tok/s | MaxMuse Spark 1.3 (Max) | Independent testIndependentArtificial Analysis ↗ | — | 1 Oct 2026 |
| 14 | gpt-oss-120b | OpenAI | 167 tok/s | Highgpt-oss-120b (High) | Independent testIndependentArtificial Analysis ↗ | — | 1 Oct 2026 |
| 15 | Mistral Medium 3.5 | Mistral AI | 165 tok/s | DefaultMistral Medium 3.5 | Independent testIndependentArtificial Analysis ↗ | — | 1 Oct 2026 |
| 16 | Nemotron 3 Super 120B A12B | NVIDIA | 163 tok/s | DefaultNemotron 3 Super 120B A12B (Reasoning) | Independent testIndependentArtificial Analysis ↗ | — | 1 Oct 2026 |
| 17 | Nemotron 3 Ultra | NVIDIA | 155 tok/s | DefaultNemotron 3 Ultra 550B A55B (Reasoning) | Independent testIndependentArtificial Analysis ↗ | — | 1 Oct 2026 |
| 18 | Muse Glimmer 30B | Meta (Meta Superintelligence Labs, Muse) | 146 tok/s | HighMuse Glimmer (High) | Independent testIndependentArtificial Analysis ↗ | — | 1 Oct 2026 |
| 19 | Claude Sonnet 5.5 | Anthropic | 139 tok/s | MaxClaude Sonnet 5.5 (Adaptive Reasoning, Max Effort, Default Fallback) | Independent testIndependentArtificial Analysis ↗ | — | 1 Oct 2026 |
| 20 | GPT-6 Luna | OpenAI | 131 tok/s | Extra highGPT-6 Luna (Xhigh) | Independent testIndependentArtificial Analysis ↗ | — | 1 Oct 2026 |
| 21 | Qwen3.6 35B A3B | Qwen (Alibaba) | 125 tok/s | DefaultQwen3.6 35B A3B (Reasoning) | Independent testIndependentArtificial Analysis ↗ | — | 1 Oct 2026 |
| 22 | GPT-5.6 Luna | OpenAI | 112 tok/s | Extra highGPT-5.6 Luna (Xhigh) | Independent testIndependentArtificial Analysis ↗ | — | 1 Oct 2026 |
| 23 | GPT-5.4 | OpenAI | 95 tok/s | Extra highGPT-5.4 (Xhigh) | Independent testIndependentArtificial Analysis ↗ | — | 1 Oct 2026 |
| 24 | MiniMax M3 | MiniMax | 91 tok/s | DefaultMiniMax-M3 | Independent testIndependentArtificial Analysis ↗ | — | 1 Oct 2026 |
| 25 | Claude Opus 5.5 | Anthropic | 90 tok/s | MaxClaude Opus 5.5 (Adaptive Reasoning, Max Effort, Default Fallback) | Independent testIndependentArtificial Analysis ↗ | — | 1 Oct 2026 |
| 26 | GPT-5.6 Terra | OpenAI | 89 tok/s | MaxGPT-5.6 Terra (Max) | Independent testIndependentArtificial Analysis ↗ | — | 1 Oct 2026 |
| 27 | DeepSeek V4 Pro (0813) | DeepSeek | 87 tok/s | MaxDeepSeek V4 Pro 0813 (Max) | Independent testIndependentArtificial Analysis ↗ | — | 1 Oct 2026 |
| 28 | Hy3 | Tencent Hunyuan | 87 tok/s | DefaultHy3 | Independent testIndependentArtificial Analysis ↗ | — | 1 Oct 2026 |
| 29 | GPT-5.5 | OpenAI | 86 tok/s | Extra highGPT-5.5 (Xhigh) | Independent testIndependentArtificial Analysis ↗ | — | 1 Oct 2026 |
| 30 | Step 5 Preview | StepFun | 85 tok/s | DefaultStep 5 Preview | Independent testIndependentArtificial Analysis ↗ | — | 1 Oct 2026 |
| 31 | GLM-5.2 | Z.ai (Zhipu) | 82 tok/s | MaxGLM-5.2 (Max) | Independent testIndependentArtificial Analysis ↗ | — | 1 Oct 2026 |
| 32 | GPT-5.6 Sol | OpenAI | 79 tok/s | MaxGPT-5.6 Sol (Max) | Independent testIndependentArtificial Analysis ↗ | — | 1 Oct 2026 |
| 33 | Claude Sonnet 5 | Anthropic | 79 tok/s | MaxClaude Sonnet 5 (Adaptive Reasoning, Max Effort) | Independent testIndependentArtificial Analysis ↗ | — | 1 Oct 2026 |
| 34 | GPT-6 Sol | OpenAI | 74 tok/s | MaxGPT-6 Sol (Max) | Independent testIndependentArtificial Analysis ↗ | — | 1 Oct 2026 |
| 35 | Grok 4.7 | SpaceXAI (formerly xAI) | 73 tok/s | HighGrok 4.7 (High) | Independent testIndependentArtificial Analysis ↗ | — | 1 Oct 2026 |
| 36 | Ministral 3 14B | Mistral AI | 73 tok/s | DefaultMinistral 3 14B | Independent testIndependentArtificial Analysis ↗ | — | 1 Oct 2026 |
| 37 | GLM-5.3 | Z.ai (Zhipu) | 69 tok/s | LowGLM-5.3 (Low) | Independent testIndependentArtificial Analysis ↗ | — | 1 Oct 2026 |
| 38 | Claude Fable 5.1 | Anthropic | 68 tok/s | MaxClaude Fable 5.1 (Adaptive Reasoning, Max Effort, Default Fallback) | Independent testIndependentArtificial Analysis ↗ | — | 1 Oct 2026 |
| 39 | Granite 4.2 8B | IBM Granite | 66 tok/s | DefaultGranite 4.2 8B | Independent testIndependentArtificial Analysis ↗ | — | 1 Oct 2026 |
| 40 | Grok 4.6 | SpaceXAI (formerly xAI) | 65 tok/s | LowGrok 4.6 (Low) | Independent testIndependentArtificial Analysis ↗ | — | 1 Oct 2026 |
| 41 | GPT-6.1 Sol | OpenAI | 64 tok/s | MaxGPT-6.1 Sol (Max) | Independent testIndependentArtificial Analysis ↗ | — | 1 Oct 2026 |
| 42 | Claude Fable 5 | Anthropic | 63 tok/s | MaxClaude Fable 5 (Adaptive Reasoning, Max Effort, Opus 4.8 Fallback) | Independent testIndependentArtificial Analysis ↗ | — | 1 Oct 2026 |
| 43 | Claude Opus 4.8 | Anthropic | 59 tok/s | MaxClaude Opus 4.8 (Adaptive Reasoning, Max Effort) | Independent testIndependentArtificial Analysis ↗ | — | 1 Oct 2026 |
| 44 | Qwen3.8-Flash-Next | Qwen (Alibaba) | 57 tok/s | DefaultQwen3.8-Flash-Next | Independent testIndependentArtificial Analysis ↗ | — | 1 Oct 2026 |
| 45 | Grok 4.5 | SpaceXAI (formerly xAI) | 55 tok/s | HighGrok 4.5 (High) | Independent testIndependentArtificial Analysis ↗ | — | 1 Oct 2026 |
| 46 | Claude Opus 5 | Anthropic | 53 tok/s | HighClaude Opus 5 (Adaptive Reasoning, High Effort) | Independent testIndependentArtificial Analysis ↗ | — | 1 Oct 2026 |
| 47 | MiMo-V2.6-Flash | Xiaomi MiMo | 51 tok/s | DefaultMiMo-V2.6-Flash | Independent testIndependentArtificial Analysis ↗ | — | 1 Oct 2026 |
| 48 | GPT-6 Astra | OpenAI | 51 tok/s | MaxGPT-6 Astra (Max) | Independent testIndependentArtificial Analysis ↗ | — | 1 Oct 2026 |
| 49 | Claude Opus 4.7 | Anthropic | 46 tok/s | MaxClaude Opus 4.7 (Adaptive Reasoning, Max Effort) | Independent testIndependentArtificial Analysis ↗ | — | 1 Oct 2026 |
| 50 | GLM-5.3-Flash | Z.ai (Zhipu) | 45 tok/s | DefaultGLM 5.3 Flash | Independent testIndependentArtificial Analysis ↗ | — | 1 Oct 2026 |
| 51 | Qwen3.8 27B | Qwen (Alibaba) | 45 tok/s | Extra highQwen3.8 27B (Xhigh) | Independent testIndependentArtificial Analysis ↗ | — | 1 Oct 2026 |
| 52 | MiMo-V2.6-Pro | Xiaomi MiMo | 42 tok/s | DefaultMiMo-V2.6-Pro | Independent testIndependentArtificial Analysis ↗ | — | 1 Oct 2026 |
| 53 | Qwen3.8 2.4T A95B (open weights) | Qwen (Alibaba) | 40 tok/s | DefaultQwen3.8 2.4T A95B | Independent testIndependentArtificial Analysis ↗ | — | 1 Oct 2026 |
| 54 | Qwen3.8 Max (0902) | Qwen (Alibaba) | 39 tok/s | DefaultQwen3.8 Max (0902) | Independent testIndependentArtificial Analysis ↗ | — | 1 Oct 2026 |
| 55 | Gemma 4 31B IT | Google (Gemini / DeepMind) | 36 tok/s | DefaultGemma 4 31B (Reasoning) | Independent testIndependentArtificial Analysis ↗ | — | 1 Oct 2026 |
| 56 | Kimi K3 | Moonshot AI (Kimi) | 34 tok/s | MaxKimi K3 (Max) | Independent testIndependentArtificial Analysis ↗ | — | 1 Oct 2026 |