AI Model Speed Rankings

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

  1. Meta Muse Spark 1.1 — Throughput: 42666.7 tps, Latency: 12200.00s
  2. xAI Grok 4.20 Multi-Agent — Throughput: 16162.7 tps, Latency: 7042.00s
  3. OpenAI o3-mini — Throughput: 13652.2 tps, Latency: 2031.00s
  4. ByteDance Seedance 2.0 — Throughput: 3178.7 tps, Latency: 0.00s
  5. ByteDance Seedance 2.0 Fast — Throughput: 3114.8 tps, Latency: 0.00s
  6. ByteDance Seedance 2.5 — Throughput: 2723.6 tps, Latency: 0.00s
  7. Meta Muse Spark 1.3 — Throughput: 2493.3 tps, Latency: 7958.00s
  8. ByteDance Seedance 2.0 Mini — Throughput: 1092.5 tps, Latency: 238079.00s
  9. Google gemini-2.5-flash — Throughput: 782.9 tps, Latency: 4830.25s
  10. Z.ai GLM 5.2 — Throughput: 426.6 tps, Latency: 1693.00s
  11. OpenAI gpt-4-gizmo-* — Throughput: 380.0 tps, Latency: 9502.50s
  12. ByteDance Doubao-Seedream-5.0-lite — Throughput: 377.0 tps, Latency: 38200.00s
  13. Meta Muse Spark 1.2 — Throughput: 346.7 tps, Latency: 14614.00s
  14. Google Gemini 3.6 Flash — Throughput: 315.8 tps, Latency: 4679.50s
  15. Google Gemini 3.5 Flash — Throughput: 257.9 tps, Latency: 8941.33s
  16. Google Gemini 3 Flash Preview — Throughput: 254.3 tps, Latency: 1937.21s
  17. Google Gemini 3.1 Flash-Lite Preview — Throughput: 246.8 tps, Latency: 0.00s
  18. Google Gemini 3.1 Pro Preview — Throughput: 236.7 tps, Latency: 11148.14s
  19. OpenAI GPT-6 Astra — Throughput: 232.5 tps, Latency: 20707.00s
  20. OpenAI GPT-5.6 Luna — Throughput: 222.8 tps, Latency: 2158.57s
  21. Google Nano Banana 2 Lite (Gemini 3.1 Flash Lite Image 🍌) — Throughput: 213.6 tps, Latency: 6808.76s
  22. Google gemini-2.5-pro — Throughput: 202.6 tps, Latency: 2089.50s
  23. OpenAI GPT-5 — Throughput: 195.2 tps, Latency: 18917.87s
  24. OpenAI GPT-6 Sol — Throughput: 194.6 tps, Latency: 1634.90s
  25. OpenAI gpt-4o-all — Throughput: 191.8 tps, Latency: 8080.00s
  26. Z.ai GLM 5.3 Flash — Throughput: 191.6 tps, Latency: 6407.00s
  27. Google Gemini 3.8 Flash — Throughput: 181.9 tps, Latency: 4462.19s
  28. DeepSeek DeepSeek V4.1 Flash — Throughput: 181.2 tps, Latency: 12716.13s
  29. Anthropic Claude Claude Haiku 5.5 — Throughput: 178.0 tps, Latency: 3875.17s
  30. xAI grok-4-fast-non-reasoning — Throughput: 172.6 tps, Latency: 666.50s
  31. Google Nano Banana (Gemini 2.5 Flash Image 🍌) — Throughput: 170.6 tps, Latency: 0.00s
  32. OpenAI chatgpt-image-latest — Throughput: 152.3 tps, Latency: 43817.00s
  33. OpenAI GPT Image 1.5 — Throughput: 137.9 tps, Latency: 48117.00s
  34. OpenAI GPT-5.4 Mini — Throughput: 134.9 tps, Latency: 11460.71s
  35. Google Gemini 3.7 Flash — Throughput: 129.4 tps, Latency: 7629.48s
  36. Anthropic Claude Claude Sonnet 5.5 — Throughput: 123.0 tps, Latency: 8594.83s
  37. Google Gemini 3.5 Flash-Lite — Throughput: 119.5 tps, Latency: 2298.83s
  38. Z.ai GLM 5 — Throughput: 116.5 tps, Latency: 10237.00s
  39. Meta Muse Wonder 1.0 — Throughput: 114.6 tps, Latency: 91059.89s
  40. OpenAI GPT-5.4 Nano — Throughput: 107.7 tps, Latency: 1654.12s
  41. OpenAI GPT-5.2 — Throughput: 107.4 tps, Latency: 664.09s
  42. Anthropic Claude Claude Opus 5.5 — Throughput: 105.6 tps, Latency: 4400.48s
  43. xAI Grok 4.20 — Throughput: 104.7 tps, Latency: 0.00s
  44. xAI grok-3 — Throughput: 97.4 tps, Latency: 0.00s
  45. Xiaomi MiMo-V2.6-Flash — Throughput: 92.3 tps, Latency: 15946.00s
  46. OpenAI gpt-4.1-mini — Throughput: 91.6 tps, Latency: 1901.00s
  47. OpenAI GPT-5.5 — Throughput: 88.0 tps, Latency: 6937.24s
  48. xAI Grok 4.3 — Throughput: 87.4 tps, Latency: 3074.17s
  49. OpenAI gpt-oss-120b — Throughput: 86.4 tps, Latency: 7326.00s
  50. xAI grok-4-fast-reasoning — Throughput: 86.3 tps, Latency: 0.00s
  51. OpenAI gpt-image-1-mini — Throughput: 82.4 tps, Latency: 12811.33s
  52. xAI grok-3-mini — Throughput: 80.0 tps, Latency: 0.00s
  53. OpenAI gpt-5-mini — Throughput: 79.3 tps, Latency: 2583.70s
  54. DeepSeek DeepSeek-V4-Pro-0813 — Throughput: 77.7 tps, Latency: 2874.40s
  55. xAI grok-4-1-fast-reasoning — Throughput: 77.6 tps, Latency: 0.00s
  56. Anthropic Claude Claude Opus 5 — Throughput: 73.3 tps, Latency: 5034.38s
  57. xAI Grok Build 0.1 — Throughput: 72.5 tps, Latency: 0.00s
  58. Google Gemini 4 Argon — Throughput: 68.5 tps, Latency: 7617.50s
  59. OpenAI gpt-5-nano — Throughput: 67.6 tps, Latency: 3154.71s
  60. xAI Grok 4.7 — Throughput: 67.4 tps, Latency: 2179.17s
  61. xAI grok-4-0709 — Throughput: 67.2 tps, Latency: 0.00s
  62. OpenAI GPT Image 2.5 Dev — Throughput: 65.8 tps, Latency: 93988.00s
  63. OpenAI gpt-4-turbo — Throughput: 64.7 tps, Latency: 0.00s
  64. ByteDance Doubao-Seed-2.0-mini 260428 — Throughput: 64.2 tps, Latency: 0.00s
  65. OpenAI GPT Image 1.0 — Throughput: 63.4 tps, Latency: 20970.12s
  66. xAI grok-4-1-fast-non-reasoning — Throughput: 61.9 tps, Latency: 0.00s
  67. Meta Muse Spark 1.2 Contributor — Throughput: 61.8 tps, Latency: 1029.00s
  68. Thinking Machines Thinking Machines: Inkling — Throughput: 61.7 tps, Latency: 0.00s
  69. xAI Grok 4.6 — Throughput: 59.6 tps, Latency: 2215.33s
  70. Google Gemini 3 Pro Preview — Throughput: 56.4 tps, Latency: 0.00s
  71. MiniMax MiniMax M2.5 — Throughput: 56.1 tps, Latency: 11687.00s
  72. Google Gemini 3.1 Flash-Lite — Throughput: 53.4 tps, Latency: 0.00s
  73. Meta Muse Spark 1.3 Contributor — Throughput: 52.4 tps, Latency: 1247.00s
  74. OpenAI GPT Image 2.5 Flare — Throughput: 51.7 tps, Latency: 24119.67s
  75. OpenAI GPT-6.1 Sol — Throughput: 51.4 tps, Latency: 1725.39s
  76. ByteDance Doubao-Seed-2.0-mini — Throughput: 49.9 tps, Latency: 0.00s
  77. Z.ai GLM 4.5 Air — Throughput: 49.8 tps, Latency: 1124.00s
  78. OpenAI gpt-3.5-turbo — Throughput: 47.4 tps, Latency: 759.00s
  79. xAI Grok 4.5 — Throughput: 45.7 tps, Latency: 3251.00s
  80. DeepSeek DeepSeek-V3.2 — Throughput: 44.4 tps, Latency: 1910.00s
  81. Z.ai GLM 5.3 — Throughput: 43.6 tps, Latency: 17144.00s
  82. MiniMax minimax-m2 — Throughput: 43.1 tps, Latency: 17669.00s
  83. ByteDance Doubao-Seed-2.1-pro — Throughput: 40.3 tps, Latency: 0.00s
  84. Alibaba Qwen3.8-Max — Throughput: 40.0 tps, Latency: 0.00s
  85. OpenAI gpt-4o-mini — Throughput: 38.7 tps, Latency: 1368.71s
  86. ByteDance Doubao-Seed-2.1-turbo — Throughput: 38.5 tps, Latency: 0.00s
  87. Anthropic Claude Claude Fable 5.1 — Throughput: 37.4 tps, Latency: 3144.00s
  88. Google Gemini Nano Banana 2.1 🍌 — Throughput: 37.1 tps, Latency: 0.00s
  89. Moonshot kimi-k2-thinking — Throughput: 37.0 tps, Latency: 0.00s
  90. Alibaba Qwen3.5 Plus 2026-02-15 — Throughput: 35.1 tps, Latency: 0.00s
  91. OpenAI gpt-4o-transcribe — Throughput: 34.9 tps, Latency: 0.00s
  92. OpenAI GPT-6 Luna — Throughput: 33.5 tps, Latency: 3701.34s
  93. Anthropic Claude Claude Opus 4.8 — Throughput: 33.3 tps, Latency: 6084.62s
  94. OpenAI gpt-3.5-turbo-0125 — Throughput: 28.5 tps, Latency: 1451.75s
  95. Anthropic Claude Claude Sonnet 5 — Throughput: 28.3 tps, Latency: 2496.14s
  96. DeepSeek DeepSeek-V4-Flash 0731 — Throughput: 28.3 tps, Latency: 2346.00s
  97. MiniMax abab5.5-chat — Throughput: 27.3 tps, Latency: 0.00s
  98. Anthropic Claude Claude Opus 4.7 — Throughput: 26.2 tps, Latency: 1922.46s
  99. OpenAI GPT-5.6 Terra — Throughput: 25.6 tps, Latency: 2694.67s
  100. DeepSeek deepseek-v3 — Throughput: 24.7 tps, Latency: 16254.00s