AI 模型速度排行榜
按实测输出吞吐(tokens/秒)与平均延迟排名。
- TypeSafe Jev 1.14 — 吞吐: 1086.8 tps, 延迟: 0.00s
- ByteDance Seedance 2.0 — 吞吐: 891.1 tps, 延迟: 364603.00s
- TypeSafe Jev 1.13 — 吞吐: 702.2 tps, 延迟: 0.00s
- ByteDance Seedance 2.5 — 吞吐: 526.4 tps, 延迟: 330005.00s
- ByteDance Doubao-Seedream-5.0-pro — 吞吐: 411.8 tps, 延迟: 53348.33s
- Google Gemini 3.6 Flash — 吞吐: 385.5 tps, 延迟: 10277.29s
- ByteDance Seedance 2.0 Mini — 吞吐: 371.8 tps, 延迟: 133826.00s
- xAI Grok 4.20 Multi-Agent — 吞吐: 370.8 tps, 延迟: 0.00s
- ByteDance Seedance 2.0 Fast — 吞吐: 365.9 tps, 延迟: 138400.00s
- Google gemini-2.5-pro — 吞吐: 264.4 tps, 延迟: 19928.97s
- Google Gemini 3 Flash Preview — 吞吐: 247.8 tps, 延迟: 6774.69s
- Anthropic Claude Claude Haiku 5.5 — 吞吐: 221.6 tps, 延迟: 1492.91s
- Google gemini-2.5-flash — 吞吐: 208.0 tps, 延迟: 6545.50s
- OpenAI gpt-5-nano — 吞吐: 204.0 tps, 延迟: 1705.29s
- Meta Muse Spark 1.3 — 吞吐: 178.7 tps, 延迟: 4749.33s
- OpenAI GPT-5 — 吞吐: 176.5 tps, 延迟: 13416.18s
- xAI grok-4-fast-non-reasoning — 吞吐: 166.7 tps, 延迟: 598.00s
- Google Gemini 3.1 Pro Preview Customtools — 吞吐: 163.3 tps, 延迟: 15981.00s
- xAI Grok Build 0.1 — 吞吐: 161.9 tps, 延迟: 0.00s
- OpenAI GPT Image 1.5 — 吞吐: 161.3 tps, 延迟: 19219.00s
- Google Nano Banana (Gemini 2.5 Flash Image 🍌) — 吞吐: 159.6 tps, 延迟: 7693.67s
- OpenAI GPT-5.4 Mini — 吞吐: 151.9 tps, 延迟: 400.08s
- Google Gemini 3.5 Flash — 吞吐: 145.0 tps, 延迟: 9102.75s
- Google Gemini 3.8 Flash — 吞吐: 144.1 tps, 延迟: 12977.88s
- OpenAI GPT Image 2 Develop — 吞吐: 144.1 tps, 延迟: 34068.64s
- OpenAI GPT-5.4 Nano — 吞吐: 143.9 tps, 延迟: 562.83s
- Anthropic Claude Claude Haiku 4.5 — 吞吐: 132.5 tps, 延迟: 2421.63s
- OpenAI GPT-5.6 Terra — 吞吐: 128.4 tps, 延迟: 2570.70s
- Anthropic Claude Claude Sonnet 5.5 — 吞吐: 128.0 tps, 延迟: 3889.14s
- DeepSeek DeepSeek V4.1 Flash — 吞吐: 126.5 tps, 延迟: 15587.98s
- OpenAI GPT-5.5 — 吞吐: 120.5 tps, 延迟: 2696.91s
- OpenAI GPT-5.6 Luna — 吞吐: 119.4 tps, 延迟: 2499.78s
- OpenAI GPT Image 2.5 Dev — 吞吐: 118.7 tps, 延迟: 35170.00s
- Anthropic Claude Claude Sonnet 4.6 — 吞吐: 116.2 tps, 延迟: 1878.56s
- OpenAI gpt-5-mini — 吞吐: 114.9 tps, 延迟: 1945.44s
- Anthropic Claude Claude Sonnet 5 — 吞吐: 112.7 tps, 延迟: 2150.91s
- Anthropic Claude Claude Opus 5.5 — 吞吐: 110.8 tps, 延迟: 3437.39s
- Z.ai GLM 5.3 Flash — 吞吐: 103.6 tps, 延迟: 1567.00s
- OpenAI GPT-5.2 — 吞吐: 103.1 tps, 延迟: 509.00s
- OpenAI GPT-6 Luna — 吞吐: 102.9 tps, 延迟: 7557.06s
- xAI Grok 4.7 — 吞吐: 100.5 tps, 延迟: 2191.59s
- xAI Grok 4.20 — 吞吐: 99.7 tps, 延迟: 0.00s
- OpenAI gpt-4o-mini-transcribe — 吞吐: 97.8 tps, 延迟: 0.00s
- OpenAI gpt-image-1-mini — 吞吐: 89.5 tps, 延迟: 14660.40s
- xAI Grok 4.6 — 吞吐: 87.2 tps, 延迟: 3872.03s
- OpenAI gpt-4o-transcribe-diarize — 吞吐: 85.1 tps, 延迟: 0.00s
- xAI grok-3 — 吞吐: 85.1 tps, 延迟: 0.00s
- xAI Grok 4.3 — 吞吐: 84.9 tps, 延迟: 2444.33s
- Alibaba Qwen3.8-Flash — 吞吐: 84.8 tps, 延迟: 0.00s
- OpenAI gpt-4o-image — 吞吐: 84.6 tps, 延迟: 5588.75s
- Google Gemini 3.1 Pro Preview — 吞吐: 83.9 tps, 延迟: 62210.39s
- xAI grok-4-1-fast-reasoning — 吞吐: 83.0 tps, 延迟: 1522.00s
- DeepSeek DeepSeek-V4-Flash 0731 — 吞吐: 82.3 tps, 延迟: 2540.25s
- Google Gemini Nano Banana 2.1 🍌 — 吞吐: 80.5 tps, 延迟: 38586.33s
- Anthropic Claude Claude Fable 5.1 — 吞吐: 78.5 tps, 延迟: 3873.77s
- xAI grok-3-mini — 吞吐: 78.3 tps, 延迟: 0.00s
- xAI grok-4-fast-reasoning — 吞吐: 77.8 tps, 延迟: 1423.50s
- xAI Grok 4.20 (Non-Reasoning) — 吞吐: 76.4 tps, 延迟: 469.50s
- MiniMax MiniMax M2.5 Lightning — 吞吐: 76.1 tps, 延迟: 0.00s
- Google Nano Banana 2 Lite (Gemini 3.1 Flash Lite Image 🍌) — 吞吐: 75.9 tps, 延迟: 19156.13s
- Z.ai GLM 5 — 吞吐: 75.5 tps, 延迟: 0.00s
- Google Gemini 4 Argon — 吞吐: 74.6 tps, 延迟: 16165.07s
- Anthropic Claude Claude Opus 4.8 — 吞吐: 74.1 tps, 延迟: 1679.83s
- OpenAI GPT Image 1.0 — 吞吐: 70.7 tps, 延迟: 78350.63s
- xAI grok-4-0709 — 吞吐: 70.5 tps, 延迟: 0.00s
- xAI Grok 4.5 — 吞吐: 69.3 tps, 延迟: 1502.11s
- Anthropic Claude Claude Opus 5 — 吞吐: 68.7 tps, 延迟: 5571.23s
- OpenAI GPT Image 2 — 吞吐: 66.4 tps, 延迟: 31448.76s
- Anthropic Claude Claude Fable 5 — 吞吐: 66.1 tps, 延迟: 2886.62s
- OpenAI GPT Image 2.5 Flare — 吞吐: 65.7 tps, 延迟: 18688.51s
- OpenAI GPT-5.3 Codex — 吞吐: 63.1 tps, 延迟: 933.69s
- OpenAI gpt-4-turbo-preview — 吞吐: 59.0 tps, 延迟: 3596.00s
- OpenAI GPT Image 2.5 Sunburst — 吞吐: 57.6 tps, 延迟: 36837.71s
- OpenAI GPT-6 Sol — 吞吐: 57.6 tps, 延迟: 1123.06s
- OpenAI GPT-6 Astra — 吞吐: 56.3 tps, 延迟: 1116.73s
- DeepSeek DeepSeek-V3.2 — 吞吐: 55.3 tps, 延迟: 1424.00s
- DeepSeek DeepSeek-V4-Pro-0813 — 吞吐: 53.6 tps, 延迟: 2033.00s
- Z.ai GLM 5.3 — 吞吐: 53.3 tps, 延迟: 0.00s
- OpenAI GPT-6.1 Sol — 吞吐: 52.8 tps, 延迟: 7509.96s
- Anthropic Claude Claude Opus 4.7 — 吞吐: 51.9 tps, 延迟: 1849.75s
- OpenAI GPT-5.6 Sol — 吞吐: 51.4 tps, 延迟: 8225.08s
- MiniMax abab5.5-chat — 吞吐: 49.6 tps, 延迟: 0.00s
- OpenAI gpt-5-pro — 吞吐: 48.4 tps, 延迟: 1528.33s
- xAI grok-4-1-fast-non-reasoning — 吞吐: 46.4 tps, 延迟: 0.00s
- OpenAI gpt-4o — 吞吐: 46.3 tps, 延迟: 4168.85s
- Moonshot moonshot-v1-32k — 吞吐: 45.9 tps, 延迟: 0.00s
- Alibaba qwen-plus-latest — 吞吐: 45.3 tps, 延迟: 1290.00s
- OpenAI gpt-4.1-mini — 吞吐: 44.4 tps, 延迟: 2537.87s
- Google Gemini 3.7 Flash — 吞吐: 43.3 tps, 延迟: 59794.96s
- Google Nano Banana 2 (Gemini 3.1 Flash Image 🍌) — 吞吐: 42.9 tps, 延迟: 35673.25s
- Moonshot moonshot-v1-8k — 吞吐: 42.6 tps, 延迟: 0.00s
- Alibaba qwen-plus-2025-07-14 — 吞吐: 40.9 tps, 延迟: 0.00s
- Moonshot Kimi K2.6 — 吞吐: 40.4 tps, 延迟: 2930.00s
- Moonshot kimi-k2-thinking — 吞吐: 38.1 tps, 延迟: 950.00s
- Google Gemini 3 Pro Preview — 吞吐: 37.6 tps, 延迟: 4092.00s
- Google Nano Banana Pro (Gemini 3 Pro Image 🍌) — 吞吐: 37.1 tps, 延迟: 32986.53s
- OpenAI GPT-Realtime-2.1 Mini — 吞吐: 36.6 tps, 延迟: 0.00s
- Anthropic Claude Claude Opus 4.6 — 吞吐: 35.9 tps, 延迟: 1826.96s
- OpenAI GPT-5.1 — 吞吐: 34.5 tps, 延迟: 1002.82s
- OpenAI GPT-5.4 — 吞吐: 33.6 tps, 延迟: 8773.95s