in the loop

GPT-5.6 Luna
#23 of 274 scored.

OpenAIJul 9, 2026Closed

Capabilities Index

Range 154.0–158.7

156.4

Context

Takes file, image, text

1.05M

Input

Per 1M tokens

$0.20

Output

Per 1M tokens

$1.20

Benchmarks

How it scores.

  • GPQA Diamond88.8%

    26th of 186

  • FrontierMath82.1%

    14th of 81

  • ARC-AGI-259.5%

    27th of 79

  • SimpleQA Verified41.0%

    43rd of 77

  • Mock AIME98.3%

    20th of 176

Everything else Epoch has for it

  • ARC-AGI88.0%
  • DTBench81.8%
  • DeepSWE67.2%
  • Surface Evolver Bench61.9%
  • FrontierMath-Tier-4-v2-Private61.0%
  • WeirdML60.9%
  • ProofBench60.0%
  • LMCA57.1%
  • Balrog45.6%
  • FrontierCode39.8%
  • Chess Puzzles36.9%
  • SimpleBench36.2%
  • Furniture Assembly17.9%
  • Mystery Game Puzzles13.0%

On the index

Around it.

  1. #20Gemini 3.8 FlashGoogle · $0.75 in156.7
  2. #21Grok 4.6xAI · $2.00 in156.4
  3. #22Qwen 3.8 MaxAlibaba156.4
  4. #23GPT-5.6 LunaOpenAI · $0.20 in156.4
  5. #24GPT-6 LunaOpenAI · $0.10 in156.3
  6. #25Claude Opus 4.7Anthropic · $5.00 in156.3
  7. #26Claude Sonnet 5Anthropic · $2.00 in156.2

Sources