AI 模型速度排行榜
按实测输出吞吐(tokens/秒)与平均延迟排名。
- xAI grok-3 — 吞吐: 15909.1 tps, 延迟: 1958.00s
- ByteDance Seedance 2.0 Mini — 吞吐: 1041.2 tps, 延迟: 145099.00s
- ByteDance Seedance 2.0 — 吞吐: 537.3 tps, 延迟: 202697.00s
- Google Gemini 3.6 Flash — 吞吐: 468.1 tps, 延迟: 10916.73s
- xAI Grok 4.20 Multi-Agent — 吞吐: 378.1 tps, 延迟: 0.00s
- Anthropic Claude Claude Sonnet 4.6 — 吞吐: 366.5 tps, 延迟: 34830.36s
- ByteDance Seedance 2.5 — 吞吐: 225.4 tps, 延迟: 225181.25s
- OpenAI GPT-5.4 Nano — 吞吐: 219.8 tps, 延迟: 2652.29s
- Google Gemini 3.1 Flash-Lite — 吞吐: 205.9 tps, 延迟: 2980.36s
- OpenAI GPT-5.4 Mini — 吞吐: 196.9 tps, 延迟: 0.00s
- Google Gemini 3 Flash Preview — 吞吐: 194.2 tps, 延迟: 18568.38s
- Google Gemini 3.1 Pro Preview — 吞吐: 191.1 tps, 延迟: 27803.76s
- Google Nano Banana (Gemini 2.5 Flash Image 🍌) — 吞吐: 190.9 tps, 延迟: 0.00s
- OpenAI GPT-5 — 吞吐: 188.9 tps, 延迟: 27103.23s
- Alibaba qwen3-next-80b-a3b-thinking — 吞吐: 187.1 tps, 延迟: 0.00s
- ByteDance Doubao-Seedream-5.0-pro — 吞吐: 179.7 tps, 延迟: 54910.83s
- Google Gemini 3 Pro Preview — 吞吐: 179.2 tps, 延迟: 16841.00s
- OpenAI GPT Image 1.5 — 吞吐: 169.1 tps, 延迟: 39953.00s
- Google Gemini 3.8 Flash — 吞吐: 168.3 tps, 延迟: 7849.02s
- OpenAI GPT Image 2.5 Dev — 吞吐: 167.4 tps, 延迟: 30445.86s
- OpenAI GPT Image 1.0 — 吞吐: 165.0 tps, 延迟: 24705.75s
- Google Nano Banana 2 Lite (Gemini 3.1 Flash Lite Image 🍌) — 吞吐: 161.1 tps, 延迟: 8628.92s
- Google gemini-2.5-flash — 吞吐: 159.9 tps, 延迟: 794.43s
- xAI Grok 4.7 — 吞吐: 159.3 tps, 延迟: 182029.35s
- OpenAI o3-mini — 吞吐: 158.7 tps, 延迟: 0.00s
- DeepSeek DeepSeek V4.1 Flash — 吞吐: 152.5 tps, 延迟: 31621.42s
- OpenAI gpt-3.5-turbo-0125 — 吞吐: 151.1 tps, 延迟: 1315.33s
- Google Gemini 3.5 Flash — 吞吐: 148.6 tps, 延迟: 12338.57s
- OpenAI gpt-5-nano — 吞吐: 147.4 tps, 延迟: 905.52s
- OpenAI GPT-6 Luna — 吞吐: 147.2 tps, 延迟: 2145.23s
- xAI Grok 4.6 — 吞吐: 143.2 tps, 延迟: 12490.75s
- OpenAI chatgpt-image-latest — 吞吐: 130.7 tps, 延迟: 35612.50s
- xAI Grok 4.3 — 吞吐: 124.2 tps, 延迟: 4010.24s
- Anthropic Claude Claude Opus 5.5 — 吞吐: 114.1 tps, 延迟: 7590.70s
- OpenAI gpt-audio-mini — 吞吐: 108.4 tps, 延迟: 0.00s
- Alibaba qwen3-30b-a3b — 吞吐: 106.8 tps, 延迟: 0.00s
- TypeSafe Jev — 吞吐: 102.7 tps, 延迟: 0.00s
- Anthropic Claude Claude Sonnet 5 — 吞吐: 100.6 tps, 延迟: 11075.34s
- DeepSeek DeepSeek-V4-Flash — 吞吐: 99.8 tps, 延迟: 1094.00s
- xAI grok-4-0709 — 吞吐: 98.1 tps, 延迟: 3361.00s
- MiniMax MiniMax M3 — 吞吐: 96.2 tps, 延迟: 9880.50s
- OpenAI gpt-4-gizmo-* — 吞吐: 95.2 tps, 延迟: 4806.69s
- xAI Grok 4.5 — 吞吐: 93.5 tps, 延迟: 13129.12s
- Alibaba Qwen3.8-Flash — 吞吐: 91.8 tps, 延迟: 0.00s
- xAI grok-4-1-fast-reasoning — 吞吐: 90.3 tps, 延迟: 0.00s
- xAI grok-3-mini — 吞吐: 90.1 tps, 延迟: 0.00s
- OpenAI gpt-audio — 吞吐: 89.4 tps, 延迟: 493.83s
- OpenAI GPT-6 Sol — 吞吐: 88.8 tps, 延迟: 2876.11s
- OpenAI GPT-5.6 Luna — 吞吐: 85.5 tps, 延迟: 1540.60s
- xAI Grok 4.20 — 吞吐: 85.5 tps, 延迟: 0.00s
- Anthropic Claude Claude Fable 5 — 吞吐: 83.6 tps, 延迟: 8844.49s
- Google Nano Banana 2 (Gemini 3.1 Flash Image 🍌) — 吞吐: 81.9 tps, 延迟: 10625.78s
- Alibaba qwen3-235b-a22b-thinking-2507 — 吞吐: 79.6 tps, 延迟: 0.00s
- OpenAI GPT-5.1 — 吞吐: 79.2 tps, 延迟: 831.05s
- xAI Grok Build 0.1 — 吞吐: 77.3 tps, 延迟: 1153.00s
- Anthropic Claude Claude Fable 5.1 — 吞吐: 76.6 tps, 延迟: 7300.67s
- Z.ai GLM 5.1 — 吞吐: 76.5 tps, 延迟: 827.50s
- Google Gemini 3.1 Pro Preview Customtools — 吞吐: 76.4 tps, 延迟: 0.00s
- OpenAI gpt-3.5-turbo — 吞吐: 75.9 tps, 延迟: 4511.50s
- OpenAI GPT-5.6 Sol — 吞吐: 74.4 tps, 延迟: 2901.00s
- Moonshot kimi-k2-thinking — 吞吐: 73.2 tps, 延迟: 402.00s
- Google gemini-2.5-pro — 吞吐: 73.1 tps, 延迟: 0.00s
- OpenAI gpt-4o-transcribe-diarize — 吞吐: 71.8 tps, 延迟: 0.00s
- OpenAI gpt-image-1-mini — 吞吐: 71.7 tps, 延迟: 14729.29s
- Z.ai GLM 4.7 — 吞吐: 71.5 tps, 延迟: 397.50s
- Anthropic Claude Claude Haiku 4.5 — 吞吐: 70.6 tps, 延迟: 14000.00s
- Alibaba qwen3-vl-plus — 吞吐: 70.1 tps, 延迟: 1830.00s
- OpenAI GPT-6 Astra — 吞吐: 69.6 tps, 延迟: 1245.60s
- Alibaba qwen3-8b — 吞吐: 68.6 tps, 延迟: 0.00s
- DeepSeek DeepSeek-V4-Pro-0813 — 吞吐: 68.4 tps, 延迟: 1169.33s
- Anthropic Claude Claude Opus 4.8 — 吞吐: 68.2 tps, 延迟: 19311.67s
- OpenAI gpt-5-codex — 吞吐: 67.3 tps, 延迟: 2948.49s
- Anthropic Claude Claude Opus 4.5 — 吞吐: 67.0 tps, 延迟: 4356.40s
- OpenAI gpt-4.1-mini — 吞吐: 65.7 tps, 延迟: 794.00s
- OpenAI GPT Image 2 Develop — 吞吐: 64.9 tps, 延迟: 35608.25s
- Z.ai GLM 5.2 — 吞吐: 61.4 tps, 延迟: 3926.50s
- OpenAI gpt-4o-2024-08-06 — 吞吐: 61.0 tps, 延迟: 0.00s
- Alibaba qwen3-32b — 吞吐: 60.5 tps, 延迟: 0.00s
- MiniMax minimax-m2 — 吞吐: 59.0 tps, 延迟: 0.00s
- OpenAI GPT-5.6 Terra — 吞吐: 58.8 tps, 延迟: 1842.12s
- DeepSeek DeepSeek-V3.2 — 吞吐: 58.8 tps, 延迟: 925.50s
- xAI Grok 4.20 (Non-Reasoning) — 吞吐: 57.1 tps, 延迟: 0.00s
- OpenAI gpt-4o-mini-transcribe — 吞吐: 55.1 tps, 延迟: 0.00s
- OpenAI gpt-5-mini — 吞吐: 54.8 tps, 延迟: 1858.44s
- Alibaba Qwen3.8-Max — 吞吐: 53.6 tps, 延迟: 0.00s
- Anthropic Claude Claude Opus 5 — 吞吐: 53.2 tps, 延迟: 7527.43s
- OpenAI GPT-5.2 — 吞吐: 52.1 tps, 延迟: 1378.06s
- Google Gemini 3.7 Flash — 吞吐: 49.5 tps, 延迟: 20785.81s
- xAI grok-4-1-fast-non-reasoning — 吞吐: 47.1 tps, 延迟: 0.00s
- Moonshot Kimi K3 — 吞吐: 47.1 tps, 延迟: 3215.50s
- OpenAI GPT Image 2.5 Flare — 吞吐: 46.3 tps, 延迟: 30549.12s
- DeepSeek DeepSeek-V4-Flash 0731 — 吞吐: 43.8 tps, 延迟: 1692.00s
- Google Gemini 3.1 Flash-Lite Preview — 吞吐: 43.5 tps, 延迟: 2579.10s
- MiniMax abab5.5-chat — 吞吐: 43.4 tps, 延迟: 0.00s
- DeepSeek deepseek-v3 — 吞吐: 42.9 tps, 延迟: 5652.84s
- Alibaba qwen3-235b-a22b — 吞吐: 41.5 tps, 延迟: 0.00s
- OpenAI GPT-5.4 — 吞吐: 41.1 tps, 延迟: 1481.55s
- Z.ai GLM 5 — 吞吐: 41.0 tps, 延迟: 704.00s
- Anthropic Claude Claude Sonnet 4.5 — 吞吐: 37.6 tps, 延迟: 5389.77s
- ByteDance Doubao-Seed-2.1-pro — 吞吐: 37.0 tps, 延迟: 0.00s