Skip to content

Terminal-Bench-Science 0.1

A benchmark of AI agents completing expert-curated research workflows across the life, physical, Earth, mathematical, and engineering sciences.

5 models

Higher is better.

  1. 1GPT-6 AstraOpenAI64.6
  2. 2Claude Sonnet 5.5Anthropic59.9
  3. 3Claude Opus 5.5Anthropic58.7
  4. 4GPT-6.1 SolOpenAI57.0
  5. 5Claude Fable 5.1Anthropic52.6
Higher is better. Bar colour: blue proprietary, orange open weights, green other.

About this benchmark

Metric
Resolution rate across 3 trials per task
Tasks
70 expert-curated scientific research workflows
Authors
Terminal-Bench-Science Team (2026)

Scores are as reported by model developers or evaluators and may use different settings (for example reasoning effort or tools). Compare with care.