SurveyBench: Analysis rating

Top 10 of 27 models · rating, higher is better · lines show the 95% range

  1. 1 Claude Fable 5.2Anthropic 1,670 (95% range 1,580–1,808)
  2. 2 GPT-6.2 AstraOpenAI 1,476 (95% range 1,402–1,580)
  3. 3 Muse WonderMeta 1,439 (95% range 1,392–1,508)
  4. 4 GPT-6.1 SolOpenAI 1,429 (95% range 1,364–1,492)
  5. 5 Claude Opus 5.5Anthropic 1,417 (95% range 1,368–1,478)
  6. 6 Qwen4-MaxAlibaba 1,381 (95% range 1,318–1,464)
  7. 7 GPT-6.2 SolOpenAI 1,362 (95% range 1,287–1,462)
  8. 8 Muse EmberMeta 1,345 (95% range 1,284–1,429)
  9. 9 Claude Haiku 5.5Anthropic 1,233 (95% range 1,194–1,291)
  10. 10 GPT-6.1 TerraOpenAI 1,217 (95% range 1,159–1,294)

Results to 18 Nov 2026 · release 3608133e8331Source: Spring Prompt · CC BY 4.0