ROASBench: Score

Top 10 of 36 models · score out of 100, higher is better · ≈ cannot be told apart from the leader

  1. 1≈ Claude Fable 5.2Anthropic 61.3
  2. 2≈ GPT-6.1 TerraOpenAI 58.4
  3. 3≈ GPT-6.2 AstraOpenAI 57.6
  4. 4≈ GPT-6 AstraOpenAI 56.6
  5. 5 Kimi K3.1 ProMoonshot AI 55.1
  6. 6 GPT-6 SolOpenAI 54.1
  7. 7 GPT-6 LunaOpenAI 53.4
  8. 8 Claude Sonnet 5.5Anthropic 52.0
  9. 9 GPT-6.2 SolOpenAI 51.8
  10. 9 Muse EmberMeta 51.8

Results to 17 Nov 2026 · release 3608133e8331Source: Spring Prompt · CC BY 4.0