Skip to content

FrontierCode 1.1 Extended

Cognition's 150-task Extended subset of the FrontierCode 1.1 software-engineering benchmark.

8 models

Higher is better.

  1. 1GPT-6 AstraOpenAI64.5
  2. 2Claude Opus 5.5Anthropic63.6
  3. 3Claude Opus 5Anthropic63.6
  4. 4Grok 4.6xAI61.3
  5. 5GPT-5.6 SolOpenAI60.6
  6. 6Claude Sonnet 5.5Anthropic59.1
  7. 7GPT-5.6 TerraOpenAI55.8
  8. 8GPT-5.6 LunaOpenAI55.1
Higher is better. Bar colour: blue proprietary, orange open weights, green other.

About this benchmark

Metric
Repository task completion with maintainer rubrics
Tasks
150 private software-engineering tasks
Authors
Cognition (2026)

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