Reaction
Gemini 4 Carbon reportedly scores 40% on FrontierCode 2.0
Cognition says its 132-task coding benchmark assesses whether maintainers would merge AI-generated code changes.
TLDR
Cognition reported a 40% score and $4.04 per task for Gemini 4 Carbon, behind Claude Opus 5.5, GPT-6 Astra and Claude Sonnet 5.5. It put Gemini 4 Argon at 26.3%, ahead of Muse Wonder 1.0 but behind Claude Haiku 5.5 and Grok 4.7. One user saw Carbon’s result as a sign Google might be back; another noted it still trailed Opus 5.5.
Combined views
—
2 Sources, first seen ago
