ResearchBenchmarksMathematics on AIME Art of Problem Solving 2025 2026 (Score)Mathematics on AIME Art of Problem Solving 2025 2026 (Score)Follow100ScoreAgentSPEX94.38495.84297.398.758Apr 14, 2026Evaluation resultsMethodMethodLinksScoreAgentSPEXModel=GPT-5, Domain=Ma...Model=GPT-5, Domain=Mathematics2026.04100CoTModel=GPT-5 (with Pyth...Model=GPT-5 (with Python), Domain=Mathematics2026.0499.6CoTModel=GPT-5 (without t...Model=GPT-5 (without tools), Domain=Mathematics2026.0494.6