CoolingModels
MathArena ranks Gemini 4 Carbon first at 85.4% expected performance
MathArena's leaderboard lists Gemini 4 Carbon (high) at 85.4% expected performance and $15.70 expected cost, ahead of Claude Fable 5.2 (max) at 82.8% and $40.20, and GPT-6.1 Sol (max) at 80.6% and $1.07. Scores are averaged across its non-deprecated competitions, so cost per problem differs sharply.
- HEAT
- OUTLETS
- 1
- FIRST SEEN
- LAST SIGNAL
Primary source
MathArenamatharena.aiCoverage · 2 articles, oldest first
Saturday, November 14, 2026
MathArenaPrimary
MathArena puts Gemini 4 Carbon first at 85.4%, at $15.70 a run