in the loop

GPT-5.6 Sol
#10 of 274 scored.

OpenAIJul 9, 2026Closed

Capabilities Index

Range 159.3–165.1

161.7

Context

Takes file, image, text

1.05M

Input

Per 1M tokens

$2.00

Output

Per 1M tokens

$10

Benchmarks

How it scores.

  • GPQA Diamond91.3%

    14th of 186

  • FrontierMath89.1%

    6th of 81

  • ARC-AGI-292.5%

    4th of 79

  • SimpleQA Verified69.7%

    7th of 77

  • Mock AIME100%

    1st of 176

Everything else Epoch has for it

  • ARC-AGI97.5%
  • DTBench93.3%
  • Surface Evolver Bench93.1%
  • WeirdML89.4%
  • ProofBench83.0%
  • FrontierMath-Tier-4-v2-Private82.9%
  • DeepSWE72.7%
  • LMCA69.7%
  • SimpleBench66.0%
  • Chess Puzzles62.1%
  • Balrog60.0%
  • Mystery Game Puzzles53.7%
  • APEX-Agents51.4%
  • FrontierCode47.5%
  • EBR-bench44.8%
  • Furniture Assembly38.1%
  • PostTrainBench36.2%
  • FrontierSWE32.2%
  • OSWorld 2.027.3%
  • MirrorCode20.0%

On the index

Around it.

  1. #7GPT-6 SolOpenAI · $2.00 in162.7
  2. #8GPT-5.5 ProOpenAI · $30 in162.1
  3. #9Claude Fable 5Anthropic · $10 in162.1
  4. #10GPT-5.6 SolOpenAI · $2.00 in161.7
  5. #11GPT-5.6 TerraOpenAI · $2.00 in159.6
  6. #12GPT-5.5OpenAI · $5.00 in159.1
  7. #13GPT-5.4 ProOpenAI · $30 in158.9

Sources