in the loop

Kimi K3
#15 of 274 scored.

MoonshotJul 16, 2026Open weights

Capabilities Index

Range 155.2–160.0

157.4

Context

Takes text, image, video

1.05M

Input

Per 1M tokens

$0.80

Output

Per 1M tokens

$14

Benchmarks

How it scores.

  • GPQA Diamond90.8%

    18th of 186

  • FrontierMath72.2%

    21st of 81

  • ARC-AGI-260.4%

    24th of 79

  • SimpleQA Verified50.6%

    24th of 77

  • Mock AIME97.2%

    26th of 176

Everything else Epoch has for it

  • Surface Evolver Bench95.0%
  • ARC-AGI94.5%
  • ProofBench87.0%
  • DTBench85.3%
  • WeirdML82.6%
  • DeepSWE68.5%
  • LMCA62.1%
  • SimpleBench52.8%
  • APEX-Agents50.6%
  • FrontierCode44.2%
  • FrontierMath-Tier-4-v2-Private39.0%
  • Chess Puzzles35.8%
  • PostTrainBench32.0%
  • FrontierSWE25.9%
  • Mystery Game Puzzles18.5%
  • Furniture Assembly5.9%

On the index

Around it.

  1. #12GPT-5.5OpenAI · $5.00 in159.1
  2. #13GPT-5.4 ProOpenAI · $30 in158.9
  3. #14Claude Opus 4.8Anthropic · $5.00 in158.2
  4. #15Kimi K3Moonshot · $0.80 in157.4
  5. #16Gemini 3.7 FlashGoogle · $0.75 in157.3
  6. #17GPT-5.4OpenAI · $2.50 in156.8
  7. #18GPT-5.3 CodexOpenAI · $1.75 in156.8

Sources