← Back to live feed1 story
Artificial Analysis published results for FrontierCode 1.1, a private evaluation created by Cognition to test if AI-generated code is accepted into real-world software repositories. Nine models from six labs scored between 43% and 52%, with Anthropic's Claude Opus 5.5 achieving the highest mark. OpenAI's GPT-6.1 Sol showed the strongest performance for its cost, scoring 49% at $0.81 per task compared to $10.15 for Claude Opus 5.5.
Sign in to suggest
Key sources
- SOURCEhuggingnewshuggingnews.com
- SOURCEhuggingnewshuggingnews.com
- SOURCEmarketbrief.now
- SOURCE@artificialanlys“Claude Opus 5.5 (max with fallback) leads at a score of 52% after passing blockers in 58% of its attempts”x.com
- SUPPORT@cognition“first benchmark to measure end-to-end code quality and mergeability: would the maintainer actually merge this PR?”x.com
- SUPPORT@teortaxestex“Good showing by V4.1 and absurdly good one from Sol 6.1, don’t look down on them”x.com
- SOURCEhuggingnewshuggingnews.com
- SOURCEmarketbrief.now