FrontierMath-Tier-4-2025-07-01-Private
Rank #13 of 51 · 27 models scored · top score 31.25
unknown
Strongest on contamination resistance, weakest on discrimination.
Weighted composite. Each signal is computed from the source-backed leaderboard below.
Does it still separate models?
How far from ceiling / clustered at the top?
Public vs held-out; training-leak risk.
Apples-to-apples, or mixed / vendor-optimized?
How old is the benchmark?
| # | Model | Score | Evidence | Measured |
|---|---|---|---|---|
| 1 | Gemini 3 Pro | 31.25 | reproduced | 2025-11-18 |
| 2 | Muse Spark | 24.33 | reproduced | 2026-04-08 |
| 3 | GLM-5.1 | 20.83 | reproduced | 2026-04-07 |
| 4 | GPT-5.1 | 20.83 | reproduced | 2025-11-13 |
| 5 | Qwen 3.6 Plus | 13.89 | reproduced | 2026-04-01 |
| 6 | Claude Sonnet 4.6 | 13.83 | reproduced | 2026-02-17 |
| 7 | Kimi K2.5 | 7 | reproduced | 2026-02-02 |
| 8 | Qwen 3.6 Max (Preview) | 6.94 | reproduced | 2026-04-20 |
| 9 | Claude Opus 4 | 6.94 | reproduced | 2025-05-22 |
| 10 | Gemini 2.5 Flash (Jun 2025) | 6.94 | reproduced | 2025-06-17 |
| 11 | GLM-4.6 | 3.55 | reproduced | 2025-09-30 |
| 12 | DeepSeek-V3.2 | 3.5 | reproduced | 2025-12-01 |
| 13 | GLM-5 | 3.5 | reproduced | 2026-02-11 |
| 14 | Grok 4 | 3.47 | reproduced | 2025-07-09 |
| 15 | Claude Haiku 4.5 | 3.47 | reproduced | 2025-10-15 |
| 16 | Qwen 3.5 Plus (hosted 397B-A17B) | 3.47 | reproduced | 2026-02-16 |
| 17 | o3 | 3.47 | reproduced | 2024-12-20 |
| 18 | Grok 3 | 0 | reproduced | 2025-02-17 |
| 19 | Kimi K2 Thinking | 0 | reproduced | 2025-11-06 |
| 20 | Claude 3.5 Sonnet | 0 | reproduced | 2024-06-20 |
| 21 | Claude 3.5 Sonnet (October 2024) | 0 | reproduced | 2024-10-22 |
| 22 | Qwen 3.5 Flash (hosted 35B-A3B) | 0 | reproduced | 2026-02-25 |
| 23 | Qwen 3.6 Flash | 0 | reproduced | 2026-04-26 |
| 24 | Qwen3-235B-A22B-Thinking (Jul 2025) | 0 | reproduced | 2025-07-25 |
| 25 | Claude Sonnet 4 | 0 | reproduced | 2025-05-22 |
| 26 | GLM-4.7 | 0 | reproduced | 2025-12-22 |
| 27 | GPT-4.1 | 0 | reproduced | 2025-04-14 |