Domains: MathScience
Skills: ReasoningCodingTool Use

Four-Factor Integrity Breakdown & Primary-Source Evidence

Privacy

Weight 40%
10/10
Access Model: Strictly Private
"We have made twelve problems public: ten from Tiers 1-3 and two from Tier 4. Unless stated otherwise, all the numbers on this hub correspond to evaluations on the private sets."

Originality

Weight 25%
10/10
Reuse Profile: All Original (0% reuse)
"We introduce FrontierMath, a benchmark of hundreds of original, exceptionally challenging mathematics problems crafted and vetted by expert mathematicians."
Primary source (Paper) ↗ Verified 2026-11-17

Freshness

Weight 20%
10/10
Initial Release: 2024-11Latest Refresh: 2026-06Rolling Updates: No
"On 2026-06-12, we released a major update, addressing errors in 42% of problems. Following this update, the full FrontierMath dataset consists of 338 problems. This is split into a base set of 295 problems, which we call Tiers 1-3, and an expansion set of 43 exceptionally difficult problems, which we call Tier 4."

Grading

Weight 15%
10/10
Integrity: Independent Offline (Leak-Verified)Sandbox: offline isolated
"For each FrontierMath question, the model needs to submit a Python function answer() that returns the answer. The answer is a Python object, often (although not always) an integer or a sympy object. Our implementation allows the model to reason and run Python code."

Frontier Model Standings on FrontierMath

Top 15 active models tracked
Rank Model Official Score Percentile
#1Claude Fable 5.2 max Anthropic100.0%99.5p
#1GPT-6.1 Sol max OpenAI100.0%99.5p
#1GPT-6.2 Astra max OpenAI100.0%99.5p
#4GPT-6 Astra max OpenAI97.6%93.1p
#5Gemini 4 Carbon high Google95.1%89.9p
#6Claude Opus 5.5 max Anthropic95.0%86.8p
#7GPT-6.1 Terra max OpenAI92.7%83.6p
#8GPT-6 Sol max OpenAI90.0%80.5p
#9Claude Fable 5.1 max Anthropic87.8%75.7p
#9Gemini 4 Hydrogen xhigh Google87.8%75.7p
#11Claude Sonnet 5.5 max Anthropic80.5%70.9p
#12Gemini 4 Argon high Google78.0%67.8p
#13Muse Wonder max Meta70.7%64.6p
#14Kimi K3.1 Pro max Kimi65.8%61.5p
#15GPT-6 Luna max OpenAI56.1%58.4p

Embeddable Audit Badge

View Markdown (.md)

Benchmark maintainers can embed this live SVG badge in GitHub READMEs or documentation:

BenchAudit FrontierMath Tier S 10/10
[![BenchAudit Tier S](https://benchaudit.org/badges/frontiermath.svg)](https://benchaudit.org/benchmarks/frontiermath/)