AIME 2026
The 2026 AIME mathematical reasoning benchmark, used to evaluate the reasoning ability of AI models on problems at the difficulty level of mathematics competitions.
Timeline (last 90 days)
Reflection AI self-reports that Beam scores 80.9 on SWE-Bench Verified, 80.1 on Terminal-Bench 2.1, and 97.8 on AIME 2026.
The Darwin series models achieved perfect scores on the AIME 2026 and HMMT 2026 mathematical reasoning benchmarks.
The RSI inference engine achieved a perfect score of 100% on both the AIME 2026 math benchmark and the HMMT 2026 math benchmark, reportedly the first perfect score on a certified math leaderboard.
VIDRAFT achieved a perfect score on the competition mathematics question sets of AIME 2026 and HMMT (February 2026).
Swift-27B's accuracy on AIME 2026 drops from 98.7% to 94.0%, a decrease of 4.6%, and the team acknowledges this is a bug in training where a critical math reasoning token was mistakenly penalized
All Facts (2)
Reflection AI self-reports that Beam scores 80.9 on SWE-Bench Verified, 80.1 on Terminal-Bench 2.1, and 97.8 on AIME 2026.
50%UnverifiedSwift-27B's accuracy on AIME 2026 drops from 98.7% to 94.0%, a decrease of 4.6%, and the team acknowledges this is a bug in training where a critical math reasoning token was mistakenly penalized
50%