Frontier Math
A benchmark maintained by Epoch AI for evaluating frontier mathematical capabilities. Problems are designed by professional mathematicians and span competition mathematics, combinatorics, number theory, and other domains. Questions are organized into difficulty tiers and are deliberately constructed to avoid solution paths already present on the internet.
Timeline (last 90 days)
OpenAI announced that Astra achieved 98% on Frontier Math Tier 4.
Previously, the pass rate of top models at the Frontier Math Tier 4 level was generally below 50%.
Frontier Math is a high-difficulty question bank designed specifically to test frontier mathematical reasoning capabilities, with Tier 4 representing the top difficulty range.
Frontier Math Tier 4 went from a top score of under 5% at launch to a top score of 98% in under 14 months, leading Epoch to conclude that the benchmark has been saturated.
Epoch AI has announced that all problems in Frontier Math Tier 4 have been solved by AI, with GPT-6 Astra solving the last unsolved problem.
Frontier Math is an evaluation benchmark maintained by Epoch AI, with problems designed by professional mathematicians that deliberately avoid solution paths already available on the internet to reduce the possibility of models answering from memorization.
The GPT-6 Astra leak claims a score of 98% on FrontierMath Tier 4.
According to Artificial Analysis data, GPT-6 Astra scored 61 on the intelligence index, tied with GPT-5.6 Sol and 5 points below Claude Fable 5.1.
Frontier Math 是由 EpochAI 创建的数学基准测试，此前 AI 模型在该测试上的通过率不足 2%
On Frontier Math, 4 of the 50 open problems have been solved, with GPT and Claude each accounting for 2.
All Facts (7)
Previously, the pass rate of top models at the Frontier Math Tier 4 level was generally below 50%.
50%UnverifiedOpenAI announced that Astra achieved 98% on Frontier Math Tier 4.
50%UnverifiedFrontier Math is a high-difficulty question bank designed specifically to test frontier mathematical reasoning capabilities, with Tier 4 representing the top difficulty range.
50%UnverifiedFrontier Math is an evaluation benchmark maintained by Epoch AI, with problems designed by professional mathematicians that deliberately avoid solution paths already available on the internet to reduce the possibility of models answering from memorization.
50%UnverifiedEpoch AI has announced that all problems in Frontier Math Tier 4 have been solved by AI, with GPT-6 Astra solving the last unsolved problem.
50%UnverifiedFrontier Math Tier 4 went from a top score of under 5% at launch to a top score of 98% in under 14 months, leading Epoch to conclude that the benchmark has been saturated.
50%UnverifiedFrontier Math 是由 EpochAI 创建的数学基准测试，此前 AI 模型在该测试上的通过率不足 2%
50%