Which AI model scores highest on FrontierMath?

GPT-5.5-Pro by OpenAI holds the best FrontierMath (Tier 4) result among tracked models, at 39.6% (released Apr 23 2026). Higher scores are better on this benchmark.

What are the top 5 models on FrontierMath?

1. GPT-5.5-Pro (OpenAI) — 39.6%; 2. GPT-5.4-Pro (OpenAI) — 38%; 3. GPT-5.5 (OpenAI) — 35.4%; 4. GPT-5.4 (OpenAI) — 27.1%; 5. Claude Opus 4.7 (Anthropic) — 22.9%.

How many models have a published FrontierMath score?

6 tracked models have a published FrontierMath (Tier 4) score. Scores are collected from published lab reports and benchmark sources, with recorded sources credited alongside the results. We do not re-run the evaluations.

Advanced math

FrontierMathTier 4

Very hard, research-level math problems. Tier 4 is the hardest — close to what professional research mathematicians tackle. Higher is better.

Tier 1–3 Tier 4 Tier 4 (v2)

Rankings

Higher is better

FrontierMath — frequently asked questions

: Very hard, research-level math problems. Tier 4 is the hardest — close to what professional research mathematicians tackle. Higher is better.
: GPT-5.5-Pro by OpenAI holds the best FrontierMath (Tier 4) result among tracked models, at 39.6% (released Apr 23 2026). Higher scores are better on this benchmark.
: 1. GPT-5.5-Pro (OpenAI) — 39.6%; 2. GPT-5.4-Pro (OpenAI) — 38%; 3. GPT-5.5 (OpenAI) — 35.4%; 4. GPT-5.4 (OpenAI) — 27.1%; 5. Claude Opus 4.7 (Anthropic) — 22.9%.
: 6 tracked models have a published FrontierMath (Tier 4) score. Scores are collected from published lab reports and benchmark sources, with recorded sources credited alongside the results. We do not re-run the evaluations.