📐 Math · source: llm-stats.com
FrontierMath
Research-grade mathematics problems crafted by professional mathematicians.
6
Models Tested
GPT-5.6 Sol
Top Model
89%
Top Score
OpenAI
Provider
| # | Model | FrontierMath |
|---|---|---|
| 1 | GPT-5.6 Sol OpenAI | 89% |
| 2 | GPT-5.6 Terra OpenAI | 84.9% |
| 3 | GPT-5.6 Luna OpenAI | 78.6% |
| 4 | GPT-5.4 OpenAI | 47.6% |
| 5 | GPT-5.2 OpenAI | 40.3% |
| 6 | GPT-5.5 OpenAI | 35.4% |
Benchmark data aggregated from llm-stats.com. See Methodology for details.