AI Model Speed Rankings

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

  1. OpenAI gpt-4-1106-preview — Throughput: 9000.0 tps, Latency: 1315.00s
  2. Google Gemini 3.5 Flash — Throughput: 2689.8 tps, Latency: 9341.18s
  3. ByteDance Seedance 2.0 Mini — Throughput: 1432.3 tps, Latency: 181600.00s
  4. Google Gemini 3.6 Flash — Throughput: 535.8 tps, Latency: 30007.00s
  5. TypeSafe Jev 1.13 — Throughput: 476.3 tps, Latency: 0.00s
  6. Xiaomi MiMo-V2.6-Pro UltraSpeed — Throughput: 469.4 tps, Latency: 4103.00s
  7. ByteDance Seedance 2.0 Fast — Throughput: 390.4 tps, Latency: 129721.00s
  8. xAI Grok 4.20 Multi-Agent — Throughput: 373.3 tps, Latency: 0.00s
  9. Google Gemini 3.1 Pro Preview — Throughput: 318.1 tps, Latency: 11081.90s
  10. OpenAI GPT-5 — Throughput: 223.7 tps, Latency: 15905.00s
  11. Anthropic Claude Claude Haiku 5.5 — Throughput: 222.9 tps, Latency: 1571.02s
  12. OpenAI GPT Image 1.5 — Throughput: 218.1 tps, Latency: 29699.33s
  13. Google gemini-2.5-flash — Throughput: 209.3 tps, Latency: 4132.00s
  14. OpenAI GPT-5.4 Mini — Throughput: 208.2 tps, Latency: 2694.87s
  15. OpenAI dall-e-3 — Throughput: 199.3 tps, Latency: 20953.00s
  16. OpenAI GPT Image 2.5 Dev — Throughput: 193.5 tps, Latency: 32299.00s
  17. Google Nano Banana 2 Lite (Gemini 3.1 Flash Lite Image 🍌) — Throughput: 185.7 tps, Latency: 9292.40s
  18. Google Gemini Nano Banana 2.1 🍌 — Throughput: 179.9 tps, Latency: 0.00s
  19. OpenAI gpt-4o-all — Throughput: 174.6 tps, Latency: 5758.50s
  20. OpenAI gpt-4-gizmo-* — Throughput: 168.6 tps, Latency: 3122.80s
  21. Google Gemini 3.8 Flash — Throughput: 166.4 tps, Latency: 4543.60s
  22. Google Gemini 3.1 Pro Preview Customtools — Throughput: 166.2 tps, Latency: 0.00s
  23. Anthropic Claude Claude Sonnet 5.5 — Throughput: 150.9 tps, Latency: 7278.88s
  24. Google Gemini 3.7 Flash — Throughput: 150.6 tps, Latency: 8575.62s
  25. ByteDance Doubao-Seedream-5.0-pro — Throughput: 150.4 tps, Latency: 77040.67s
  26. OpenAI gpt-audio-mini — Throughput: 150.0 tps, Latency: 384.09s
  27. xAI Grok 4.3 — Throughput: 148.4 tps, Latency: 2570.27s
  28. OpenAI chatgpt-image-latest — Throughput: 146.6 tps, Latency: 49308.00s
  29. OpenAI GPT-6 Sol — Throughput: 138.8 tps, Latency: 740.00s
  30. DeepSeek DeepSeek V4.1 Flash — Throughput: 132.8 tps, Latency: 4406.61s
  31. xAI grok-4-1-fast-reasoning — Throughput: 125.7 tps, Latency: 0.00s
  32. OpenAI GPT-5.1 — Throughput: 124.0 tps, Latency: 611.00s
  33. OpenAI GPT Image 2 Develop — Throughput: 123.2 tps, Latency: 22863.14s
  34. OpenAI gpt-4.1-nano — Throughput: 118.0 tps, Latency: 442.65s
  35. Alibaba qwen3-coder-flash — Throughput: 115.8 tps, Latency: 1616.00s
  36. xAI grok-3-mini — Throughput: 113.5 tps, Latency: 0.00s
  37. Google Nano Banana (Gemini 2.5 Flash Image 🍌) — Throughput: 112.5 tps, Latency: 12854.00s
  38. ByteDance Doubao-Seed-2.0-mini 260428 — Throughput: 110.4 tps, Latency: 0.00s
  39. xAI grok-3 — Throughput: 108.0 tps, Latency: 0.00s
  40. xAI Grok 4.20 — Throughput: 107.5 tps, Latency: 742.00s
  41. Anthropic Claude Claude Opus 5.5 — Throughput: 103.8 tps, Latency: 5380.60s
  42. MiniMax minimax-m2.1 — Throughput: 103.3 tps, Latency: 2603.00s
  43. OpenAI GPT-5.4 Nano — Throughput: 103.1 tps, Latency: 831.50s
  44. Alibaba Qwen3.7 Max — Throughput: 102.8 tps, Latency: 3830.00s
  45. Meta Muse Wonder 1.0 — Throughput: 99.6 tps, Latency: 0.00s
  46. OpenAI GPT-5.6 Terra — Throughput: 99.0 tps, Latency: 0.00s
  47. xAI Grok Build 0.1 — Throughput: 98.9 tps, Latency: 0.00s
  48. OpenAI gpt-5-nano — Throughput: 96.4 tps, Latency: 1278.88s
  49. MiniMax MiniMax M2.5 — Throughput: 95.7 tps, Latency: 6362.00s
  50. Moonshot Kimi K2.6 — Throughput: 92.4 tps, Latency: 9948.00s
  51. DeepSeek DeepSeek-V4-Flash — Throughput: 92.2 tps, Latency: 4560.80s
  52. Z.ai GLM 5 — Throughput: 90.1 tps, Latency: 6745.25s
  53. OpenAI GPT-5.6 Sol — Throughput: 90.0 tps, Latency: 1374.78s
  54. DeepSeek DeepSeek-V4-Pro-0813 — Throughput: 89.3 tps, Latency: 4107.17s
  55. Google Gemini 4 Argon — Throughput: 89.2 tps, Latency: 5795.74s
  56. Google Nano Banana 2 (Gemini 3.1 Flash Image 🍌) — Throughput: 86.0 tps, Latency: 10938.86s
  57. OpenAI gpt-3.5-turbo — Throughput: 85.3 tps, Latency: 0.00s
  58. Z.ai GLM 5 Turbo — Throughput: 84.7 tps, Latency: 12867.00s
  59. OpenAI gpt-5-mini — Throughput: 83.8 tps, Latency: 4559.00s
  60. Anthropic Claude Claude Opus 5 — Throughput: 83.2 tps, Latency: 3594.97s
  61. Google gemini-2.5-pro — Throughput: 81.9 tps, Latency: 2084.50s
  62. OpenAI gpt-image-1-mini — Throughput: 81.8 tps, Latency: 12904.00s
  63. OpenAI gpt-4.1-mini — Throughput: 81.1 tps, Latency: 0.00s
  64. Alibaba Qwen3.8-Flash — Throughput: 78.8 tps, Latency: 29325.50s
  65. Z.ai GLM 5.2 — Throughput: 78.5 tps, Latency: 11716.08s
  66. Anthropic Claude Claude Sonnet 5 — Throughput: 76.9 tps, Latency: 2883.00s
  67. Google Gemini 3.5 Flash-Lite — Throughput: 75.5 tps, Latency: 1368.62s
  68. Anthropic Claude Claude Fable 5.1 — Throughput: 74.4 tps, Latency: 3019.50s
  69. OpenAI GPT-5.5 — Throughput: 73.8 tps, Latency: 1236.00s
  70. OpenAI gpt-4o-transcribe-diarize — Throughput: 71.8 tps, Latency: 0.00s
  71. Anthropic Claude Claude Opus 4.7 — Throughput: 70.4 tps, Latency: 1632.10s
  72. xAI Grok 4.7 — Throughput: 66.1 tps, Latency: 11926.00s
  73. OpenAI gpt-3.5-turbo-16k — Throughput: 63.9 tps, Latency: 5737.00s
  74. Z.ai GLM 4.5 Air — Throughput: 61.7 tps, Latency: 0.00s
  75. OpenAI GPT Image 1.0 — Throughput: 61.2 tps, Latency: 25870.00s
  76. DeepSeek DeepSeek-V4-Flash 0731 — Throughput: 60.5 tps, Latency: 5184.12s
  77. xAI grok-4-fast-reasoning — Throughput: 60.4 tps, Latency: 0.00s
  78. Meta Muse Image 1.0 — Throughput: 59.8 tps, Latency: 25217.00s
  79. Alibaba Qwen3.5 Plus 2026-02-15 — Throughput: 58.8 tps, Latency: 1520.00s
  80. OpenAI GPT-6 Luna — Throughput: 58.0 tps, Latency: 2360.15s
  81. OpenAI GPT-5.4 — Throughput: 56.9 tps, Latency: 2554.60s
  82. OpenAI gpt-4.1 — Throughput: 56.7 tps, Latency: 2489.00s
  83. Anthropic Claude Claude Opus 4.8 — Throughput: 56.1 tps, Latency: 2292.64s
  84. OpenAI GPT-6.1 Sol — Throughput: 56.1 tps, Latency: 2684.31s
  85. xAI Grok 4.6 — Throughput: 55.5 tps, Latency: 1334.75s
  86. MiniMax MiniMax M2.5 Lightning — Throughput: 53.3 tps, Latency: 0.00s
  87. OpenAI gpt-4o-mini-transcribe — Throughput: 52.3 tps, Latency: 0.00s
  88. xAI grok-4-1-fast-non-reasoning — Throughput: 51.8 tps, Latency: 0.00s
  89. xAI Grok 4.5 — Throughput: 51.3 tps, Latency: 1823.90s
  90. Alibaba Qwen3.6 Plus — Throughput: 51.0 tps, Latency: 0.00s
  91. xAI grok-4-0709 — Throughput: 50.6 tps, Latency: 18441.00s
  92. ByteDance Doubao-Seed-2.0-mini — Throughput: 50.2 tps, Latency: 0.00s
  93. Anthropic Claude Claude Fable 5 — Throughput: 49.8 tps, Latency: 2880.61s
  94. Google Gemini 3 Pro Preview — Throughput: 49.4 tps, Latency: 0.00s
  95. Alibaba Qwen3.8-Max — Throughput: 47.8 tps, Latency: 0.00s
  96. OpenAI GPT-5.6 Luna — Throughput: 47.8 tps, Latency: 2908.48s
  97. OpenAI gpt-oss-120b — Throughput: 46.2 tps, Latency: 299489.29s
  98. Z.ai GLM 5.3 Flash — Throughput: 45.3 tps, Latency: 11199.12s
  99. Google Gemini 3.1 Flash-Lite — Throughput: 44.4 tps, Latency: 0.00s
  100. OpenAI GPT Image 2.5 Sunburst — Throughput: 43.2 tps, Latency: 26957.64s