Listed speed

Fastest listed output speed

Indexable models with a published output speed, sorted fast to slow. These are catalog or benchmark medians, not a measurement of your own traffic.

272 models with this published figure · Artificial Analysis median output speed, where published.

Markdown snapshot for answer engines.

Data: Artificial Analysis

  1. 1Meta logoMuse EmberMeta452.2 t/s
  2. 2Meta logoMuse Ember (Xhigh)Meta418.4 t/svs Muse Ember
  3. 3Liquid AI logoLFM3-VL-2BLiquid AI355.7 t/svs Muse Ember
  4. 4Liquid AI logoLFM3-VL-2BLiquid AI339.6 t/svs Muse Ember
  5. 5Google logoGemini 3.1 Flash-LiteGoogle320.4 t/svs Muse Ember
  6. 6NVIDIA logoNemotron 3.5 Super VLNVIDIA310.9 t/svs Muse Ember
  7. 7Google logoGemini 3.5 Flash-LiteGoogle307 t/svs Muse Ember
  8. 8Reflection logoBeam 1.0Reflection297.6 t/svs Muse Ember
  9. 9Thinking Machines logoInkling 2 MiniThinking Machines279.9 t/svs Muse Ember
  10. 10Thinking Machines logoInkling 2 MiniThinking Machines279.9 t/svs Muse Ember
  11. 11Google logoGemini 2.5 Flash LiteGoogle278.5 t/svs Muse Ember
  12. 12Institute of Foundation Models logoK2 Horizon 375B A23BInstitute of Foundation Models277.7 t/svs Muse Ember
  13. 13NVIDIA logoNemotron 3.5 LightningNVIDIA275.5 t/svs Muse Ember
  14. 14Anthropic logoClaude Haiku 5.5Anthropic257.3 t/svs Muse Ember
  15. 15Liquid AI logoLFM3-1.7BLiquid AI255.3 t/svs Muse Ember
  16. 16Anthropic logoClaude Haiku 5.5 (Xhigh)Anthropic240.6 t/svs Muse Ember
  17. 17Cohere logoNorth Mini CodeCohere230 t/svs Muse Ember
  18. 18Anthropic logoClaude Haiku 5.5 (High)Anthropic227.5 t/svs Muse Ember
  19. 19DeepSeek logoDeepSeek V4.1 FlashDeepSeek222.2 t/svs Muse Ember
  20. 20Anthropic logoClaude Haiku 5.5 (Medium)Anthropic220.1 t/svs Muse Ember
  21. 21Anthropic logoClaude Haiku 5.5 (Low)Anthropic217.4 t/svs Muse Ember
  22. 22OpenAI logoo3-mini (High)OpenAI217.4 t/svs Muse Ember
  23. 23Meta logoMuse Spark 1.2Meta211.3 t/svs Muse Ember
  24. 24Meta logoMuse Spark 1.1Meta207.3 t/svs Muse Ember
  25. 25NVIDIA logoNemotron 3 Ultra 550B A55B (Reasoning)NVIDIA207.2 t/svs Muse Ember
  26. 26Cohere logoCommand A+Cohere204.4 t/svs Muse Ember
  27. 27OpenAI logogpt-oss-120bOpenAI191.2 t/svs Muse Ember
  28. 28OpenAI logoo3-miniOpenAI186.3 t/svs Muse Ember
  29. 29OpenAI logoGPT-5.4 MiniOpenAI177.3 t/svs Muse Ember
  30. 30Meta logoMuse Spark OpenMeta175.6 t/svs Muse Ember
  31. 31Google logoGemini 3.5 FlashGoogle173.7 t/svs Muse Ember
  32. 32Sapiens AI logoAgnes 2.5 Pro ThetaSapiens AI172 t/svs Muse Ember
  33. 33Google logoGemini 4 Hydrogen (Medium)Google171 t/svs Muse Ember
  34. 34Google logoGemini 2.5 FlashGoogle170.9 t/svs Muse Ember
  35. 35OpenAI logoGPT-5 NanoOpenAI169.9 t/svs Muse Ember
  36. 36DeepSeek logoDeepSeek V4.1 Flash 1017 (High)DeepSeek169.3 t/svs Muse Ember
  37. 37InclusionAI logoLing 3.0 TinyInclusionAI165.7 t/svs Muse Ember
  38. 38OpenAI logoGPT-4.1 nanoOpenAI165.6 t/svs Muse Ember
  39. 39Google logoGemini 3.6 FlashGoogle164.3 t/svs Muse Ember
  40. 40DeepSeek logoDeepSeek V4.1 Flash 1017DeepSeek163.7 t/svs Muse Ember

Questions

How is the “Fastest listed output speed” list ranked?
Indexable models with a published output speed, sorted fast to slow. These are catalog or benchmark medians, not a measurement of your own traffic. 272 models currently have this published figure.
Where does the data come from?
Artificial Analysis median output speed, where published. Empty fields mean the source has not published that number.
Is this the best model overall?
No. Each list is sorted by a single published signal. Maker Crown does not invent a combined score or a “best AI model” award.