Artificial Analysis MATH-500 (AA MATH-500)
We show this table for reference; we do not rank on it.
An independently evaluated MATH-500 result from Artificial Analysis.
Benchmark score on AA MATH-500 — September 27, 2026
We compile the AA MATH-500 rows from secondary reports. GPT-5 (high) leads the table at 99.4%, followed by o3 (99.2%) and GPT-5 (medium) (99.1%). We do not use these results to rank models overall.
GPT-5 (high)
OpenAI
o3
OpenAI
GPT-5 (medium)
OpenAI
3 modelsMathematicsCurrentDisplay onlyUpdated September 27, 2026
Benchmark score table (3 models)
ScoreAmong the reported AA MATH-500 rows, GPT-5 (high) is first at 99.4%. The third row is 0.3 points behind. The broader top-10 range is 0.3 points, so many of the published results sit in a relatively narrow band.
3 models have been evaluated on AA MATH-500. The benchmark falls in the Mathematics category. AA MATH-500 is currently displayed for reference but excluded from the scoring formula, so it does not directly affect overall rankings.
About AA MATH-500
Year
2026
Tasks
500 competition mathematics problems
Format
Accuracy
Difficulty
High school to undergraduate mathematics
Stored separately from the core MATH-500 lane because Artificial Analysis controls the evaluation configuration.
Freshness and provenance
Version
AA MATH-500 2026
Refresh cadence
Quarterly
Staleness state
Current
Question availability
Public benchmark set
BenchLM uses freshness metadata to decide whether a benchmark should still be treated as a strong differentiator, a benchmark to watch, or a display-only reference. For the full scoring policy, see the BenchLM methodology page.
Questions
What does AA MATH-500 measure?
An independently evaluated MATH-500 result from Artificial Analysis.
Which model scores highest on AA MATH-500?
GPT-5 (high) by OpenAI currently leads with a score of 99.4% on AA MATH-500.
How many models are evaluated on AA MATH-500?
3 AI models have been evaluated on AA MATH-500 on BenchLM.
Compare top models on AA MATH-500
Know when it’s worth switching models
The model to choose, the cheaper alternative, and the release we would wait on.
Read a sample issueJoin 2,000+ readers.
One email each week. Unsubscribe anytime.