Frontier Math Benchmark Leaderboard | Phaseo

Frontier Math

Type: percentageMath

Recorded Results

14

Average Score

33.44%

Score Range

9.60% - 52.40%

Leading Model (lowest score)

9.60% - GPT 5 Nano

Scores Over Time

Individual benchmark scores plotted by date.

Models Using This Benchmark

Organisation	Model	Reported	Top Score	Info	Self Reported	Source
OpenAI	GPT 5 Nano	07 Aug 2025	9.60%	With Thinking, With Python, Pass @ 1	Yes	Source
OpenAI	GPT 5 Mini	07 Aug 2025	22.10%	With Thinking, With Python, Pass @ 1	Yes	Source
OpenAI	GPT 5	07 Aug 2025	26.30%	With Thinking, With Python, Pass @ 1	Yes	Source
OpenAI	GPT 5.2	11 Dec 2025	40.30%	Tier 1-3 With Python	Yes	Source
OpenAI	GPT 5.4	05 Mar 2026	47.60%	Tier 1-3	Yes	Source
OpenAI	GPT 5.4 Pro	05 Mar 2026	50%	Tier 1-3	Yes	Source
OpenAI	GPT 5.5	23 Apr 2026	51.70%	Tier 1-3	Yes	Source
OpenAI	GPT 5.5 Pro	23 Apr 2026	52.40%	Tier 1-3	Yes	Source