AI Model Speed Rankings

Models ranked by measured output throughput (tokens per second) and average latency.

  1. Meta Muse Wonder 1.0 — Throughput: 1122.2 tps, Latency: 2127.67s
  2. ByteDance Seedance 2.0 — Throughput: 912.0 tps, Latency: 664139.00s
  3. OpenAI o4-mini — Throughput: 866.3 tps, Latency: 3852.00s
  4. ByteDance Seedance 2.5 — Throughput: 849.8 tps, Latency: 317045.00s
  5. TypeSafe Jev 1.13 — Throughput: 756.0 tps, Latency: 0.00s
  6. ByteDance Doubao-Seedream-5.0-lite — Throughput: 742.5 tps, Latency: 24674.75s
  7. TypeSafe Jev 1.14 — Throughput: 625.0 tps, Latency: 0.00s
  8. xAI grok-4-fast-reasoning — Throughput: 541.2 tps, Latency: 1316.80s
  9. xAI Grok 4.20 Multi-Agent — Throughput: 494.3 tps, Latency: 0.00s
  10. ByteDance Seedance 2.0 Mini — Throughput: 469.8 tps, Latency: 86410.00s
  11. OpenAI GPT-5 — Throughput: 435.4 tps, Latency: 17076.38s
  12. ByteDance Seedance 2.0 Fast — Throughput: 420.4 tps, Latency: 120438.00s
  13. Google Gemini 3.6 Flash — Throughput: 260.9 tps, Latency: 2037.50s
  14. Google Gemini 3 Flash Preview — Throughput: 238.1 tps, Latency: 7543.80s
  15. Anthropic Claude Claude Sonnet 5.5 — Throughput: 220.7 tps, Latency: 5372.78s
  16. Google Gemini 3.5 Flash — Throughput: 216.8 tps, Latency: 8374.33s
  17. OpenAI GPT Image 1.5 — Throughput: 216.5 tps, Latency: 30640.00s
  18. Google Nano Banana 2 Lite (Gemini 3.1 Flash Lite Image 🍌) — Throughput: 209.0 tps, Latency: 7370.67s
  19. Google Gemini 3.1 Pro Preview — Throughput: 209.0 tps, Latency: 8919.97s
  20. OpenAI GPT-5.4 Mini — Throughput: 202.5 tps, Latency: 503.38s
  21. DeepSeek DeepSeek-V4-Flash — Throughput: 198.4 tps, Latency: 6069.00s
  22. Google Gemini Nano Banana 2.1 🍌 — Throughput: 188.6 tps, Latency: 13424.33s
  23. OpenAI gpt-5-nano — Throughput: 188.4 tps, Latency: 366.08s
  24. OpenAI dall-e-2 — Throughput: 176.2 tps, Latency: 23701.00s
  25. Google Gemini 3.1 Flash-Lite Preview — Throughput: 175.5 tps, Latency: 4335.00s
  26. OpenAI GPT Image 2.5 Dev — Throughput: 172.2 tps, Latency: 27954.15s
  27. OpenAI GPT-5.6 Terra — Throughput: 172.1 tps, Latency: 1421.15s
  28. Google Gemini 3.1 Pro Preview Customtools — Throughput: 167.7 tps, Latency: 0.00s
  29. OpenAI GPT Image 1.0 — Throughput: 158.3 tps, Latency: 39491.75s
  30. OpenAI GPT-5.1 — Throughput: 153.0 tps, Latency: 476.82s
  31. Anthropic Claude Claude Sonnet 4.5 — Throughput: 148.6 tps, Latency: 2728.83s
  32. Anthropic Claude Claude Haiku 5.5 — Throughput: 139.9 tps, Latency: 1335.51s
  33. Google gemini-2.5-flash — Throughput: 138.4 tps, Latency: 7541.46s
  34. xAI Grok 4.20 — Throughput: 116.6 tps, Latency: 0.00s
  35. Z.ai GLM 5.3 Flash — Throughput: 114.2 tps, Latency: 2233.67s
  36. OpenAI dall-e-3 — Throughput: 112.6 tps, Latency: 37082.00s
  37. Google Gemini 3.7 Flash — Throughput: 107.9 tps, Latency: 12421.62s
  38. DeepSeek DeepSeek-V4-Flash 0731 — Throughput: 105.6 tps, Latency: 1544.00s
  39. OpenAI gpt-image-1-mini — Throughput: 102.9 tps, Latency: 15236.00s
  40. xAI Grok Build 0.1 — Throughput: 102.2 tps, Latency: 0.00s
  41. OpenAI GPT-6 Sol — Throughput: 101.4 tps, Latency: 1124.00s
  42. OpenAI GPT-5.5 — Throughput: 100.5 tps, Latency: 3440.65s
  43. OpenAI gpt-3.5-turbo — Throughput: 99.2 tps, Latency: 2227.00s
  44. DeepSeek DeepSeek V4.1 Flash — Throughput: 95.8 tps, Latency: 1652.00s
  45. Anthropic Claude Claude Opus 5.5 — Throughput: 94.5 tps, Latency: 3511.97s
  46. ByteDance Doubao-Seedream-5.0-pro — Throughput: 92.9 tps, Latency: 66397.50s
  47. OpenAI GPT-5.4 Nano — Throughput: 91.2 tps, Latency: 527.25s
  48. Google Gemini 3.8 Flash — Throughput: 87.1 tps, Latency: 15971.67s
  49. Google Gemini 4 Argon — Throughput: 86.5 tps, Latency: 25167.33s
  50. xAI grok-3 — Throughput: 86.0 tps, Latency: 0.00s
  51. Alibaba Qwen3.8-Flash — Throughput: 85.5 tps, Latency: 0.00s
  52. OpenAI GPT Image 2.5 Flare — Throughput: 85.4 tps, Latency: 14474.05s
  53. Anthropic Claude Claude Opus 4.7 — Throughput: 81.2 tps, Latency: 2472.68s
  54. Google gemini-2.5-pro — Throughput: 80.9 tps, Latency: 10645.28s
  55. xAI Grok 4.3 — Throughput: 80.3 tps, Latency: 1800.67s
  56. Anthropic Claude Claude Opus 5 — Throughput: 79.7 tps, Latency: 2361.18s
  57. xAI Grok 4.20 (Non-Reasoning) — Throughput: 79.1 tps, Latency: 0.00s
  58. OpenAI GPT-6 Astra — Throughput: 79.0 tps, Latency: 673.21s
  59. xAI Grok 4.6 — Throughput: 74.7 tps, Latency: 4582.21s
  60. OpenAI GPT-5.1 Codex Max — Throughput: 73.6 tps, Latency: 2504.00s
  61. OpenAI GPT Image 2 Develop — Throughput: 73.2 tps, Latency: 45614.79s
  62. OpenAI GPT-5.4 — Throughput: 71.2 tps, Latency: 4097.58s
  63. OpenAI GPT-5.6 Luna — Throughput: 70.0 tps, Latency: 1026.01s
  64. Anthropic Claude Claude Opus 4.8 — Throughput: 69.6 tps, Latency: 1822.89s
  65. xAI grok-4-fast-non-reasoning — Throughput: 69.0 tps, Latency: 522.00s
  66. xAI grok-3-mini — Throughput: 67.4 tps, Latency: 0.00s
  67. xAI grok-4-0709 — Throughput: 66.7 tps, Latency: 1028.00s
  68. xAI grok-4-1-fast-reasoning — Throughput: 65.3 tps, Latency: 0.00s
  69. Anthropic Claude Claude Sonnet 5 — Throughput: 64.2 tps, Latency: 1435.53s
  70. OpenAI gpt-audio-mini — Throughput: 63.8 tps, Latency: 0.00s
  71. xAI Grok 4.7 — Throughput: 63.8 tps, Latency: 2695.60s
  72. OpenAI GPT-5.2 — Throughput: 63.3 tps, Latency: 1143.30s
  73. xAI Grok 4.5 — Throughput: 62.4 tps, Latency: 3643.10s
  74. MiniMax MiniMax M3 — Throughput: 61.1 tps, Latency: 0.00s
  75. Anthropic Claude Claude Sonnet 4.6 — Throughput: 60.4 tps, Latency: 1170.33s
  76. OpenAI GPT-6.1 Sol — Throughput: 60.2 tps, Latency: 2462.59s
  77. ByteDance Doubao-Seed-2.1-turbo — Throughput: 58.5 tps, Latency: 69468.00s
  78. Z.ai GLM 5.3 — Throughput: 55.2 tps, Latency: 1418.33s
  79. DeepSeek DeepSeek-V4-Pro-0813 — Throughput: 52.8 tps, Latency: 1869.00s
  80. OpenAI GPT Image 2.5 Sunburst — Throughput: 49.7 tps, Latency: 28537.38s
  81. Moonshot kimi-k2-0905-preview — Throughput: 47.1 tps, Latency: 1822.00s
  82. Alibaba Qwen3.7 Max — Throughput: 45.5 tps, Latency: 0.00s
  83. Moonshot Kimi K2.6 — Throughput: 45.4 tps, Latency: 17005.00s
  84. xAI grok-4-1-fast-non-reasoning — Throughput: 41.2 tps, Latency: 1065.17s
  85. Google Gemini 3 Pro Preview — Throughput: 41.2 tps, Latency: 0.00s
  86. OpenAI gpt-4o-transcribe — Throughput: 40.6 tps, Latency: 0.00s
  87. OpenAI GPT-Realtime — Throughput: 40.2 tps, Latency: 0.00s
  88. Alibaba qwen-plus-2025-07-14 — Throughput: 39.7 tps, Latency: 0.00s
  89. Alibaba Qwen3.6 Plus — Throughput: 39.7 tps, Latency: 0.00s
  90. DeepSeek deepseek-v3.1-terminus — Throughput: 39.6 tps, Latency: 0.00s
  91. Z.ai GLM 5 — Throughput: 39.3 tps, Latency: 0.00s
  92. Moonshot moonshot-v1-128k-vision-preview — Throughput: 38.7 tps, Latency: 0.00s
  93. OpenAI GPT-6 Luna — Throughput: 38.4 tps, Latency: 3898.41s
  94. Moonshot moonshot-v1-32k-vision-preview — Throughput: 37.9 tps, Latency: 0.00s
  95. Anthropic Claude Claude Opus 4.5 — Throughput: 37.5 tps, Latency: 1710.00s
  96. Moonshot moonshot-v1-8k-vision-preview — Throughput: 36.9 tps, Latency: 0.00s
  97. ByteDance Doubao-Seed-2.1-pro — Throughput: 36.7 tps, Latency: 0.00s
  98. Moonshot Kimi K3 — Throughput: 35.7 tps, Latency: 1363.00s
  99. MiniMax MiniMax-M2.7 — Throughput: 35.5 tps, Latency: 0.00s
  100. Anthropic Claude Claude Opus 4.6 — Throughput: 35.1 tps, Latency: 1739.36s