Frontiermath tier 4 v2 — Epoch AI
Frontiermath tier 4 v2 — Epoch AI is run by Epoch AI. It has ranked 62 models, of which the catalogue holds 54, scoring from 0 to 97.6.
measured by Epoch AI · the board itself
| Place | Model | Metric | Score |
|---|---|---|---|
| 1stof 62 | GPT 6 Astra | mean_score (high) | 97.6 |
| 5thof 62 | Claude Fable 5 | mean_score (max) | 90.2 |
| 6thof 62 | Claude Fable 5.1 | mean_score (max) | 87.805 |
| 9thof 62 | GPT-5.6 Sol | mean_score (max) | 82.927 |
| 11thof 62 | GPT-5.5 Pro | mean_score (xhigh) | 78.049 |
| 13thof 62 | Claude Opus 5 | mean_score (max) | 73.171 |
| 14thof 62 | GPT-5.5 | mean_score (xhigh) | 72.5 |
| 15thof 62 | GPT-5.6 Terra | mean_score (max) | 70.732 |
| 16thof 62 | GPT-5.6 Luna | mean_score (max) | 60.976 |
| 17thof 62 | GPT-5.4 Pro | mean_score (xhigh) | 58.537 |
| 18thof 62 | Claude Opus 4.8 | mean_score (max) | 56.098 |
| 19thof 62 | GPT-5.4 | mean_score (xhigh) | 49 |
| 20thof 62 | Qwen 3.8 Max | mean_score (xhigh) | 46.341 |
| 21stof 62 | GPT-5.2 Pro | mean_score (xhigh) | 46 |
| 22ndof 62 | Kimi K3 | mean_score (max) | 39.024 |
| 23rdof 62 | Gemini 3.7 Flash | mean_score (high) | 36.585 |
| 25thof 62 | Qwen3.7 Max | mean_score (max) | 34.146 |
| 26thof 62 | Grok 4.6 | mean_score (xhigh) | 31.707 |
| 27thof 62 | Claude Opus 4.7 | mean_score (max) | 31.707 |
| 28thof 62 | GPT-5.2 | mean_score (xhigh) | 31.7 |
| 29thof 62 | GLM-5.3 | mean_score (max) | 29.268 |
| 30thof 62 | Claude Sonnet 5 | mean_score (max) | 29.268 |
| 31stof 62 | GLM 5.2 | mean_score (max) | 29.268 |
| 32ndof 62 | DeepSeek V4 Pro 0813 | mean_score (max) | 26.829 |
| 33rdof 62 | Gemini 3.1 Pro Preview | mean_score | 26.829 |
| 34thof 62 | Claude Opus 4.6 | mean_score (max) | 26.829 |
| 35thof 62 | Gemini 3.5 Flash | mean_score (high) | 26.829 |
| 36thof 62 | Kimi K2.6 | mean_score | 25.641 |
| 37thof 62 | DeepSeek V4 Flash (0731) | mean_score (max) | 24.39 |
| 38thof 62 | Grok 4.5 | mean_score (high) | 24.39 |
| 39thof 62 | Gemini 3.8 Flash | mean_score (high) | 21.951 |
| 40thof 62 | Gemini 3.6 Flash | mean_score (high) | 21.951 |
| 41stof 62 | GPT-5 | mean_score (high) | 21.951 |
| 42ndof 62 | GPT-5 Pro | mean_score (high) | 19.512 |
| 43rdof 62 | GLM 5.3 Flash | mean_score (max) | 17.073 |
| 44thof 62 | Inkling Small | mean_score (xhigh) | 17.073 |
| 45thof 62 | Grok 4.20 | mean_score | 17.073 |
| 46thof 62 | Gemini 3 Flash Preview | mean_score | 17.073 |
| 47thof 62 | Grok 4.3 | mean_score (high) | 14.634 |
| 48thof 62 | Kimi K2.7 Code | mean_score | 12.195 |
| 49thof 62 | GPT-5 mini | mean_score (high) | 12.195 |
| 50thof 62 | GPT-5.4 nano | mean_score (high) | 12.195 |
| 51stof 62 | GPT-5.4 mini | mean_score (xhigh) | 9.756 |
| 52ndof 62 | Inkling | mean_score (xhigh) | 4.878 |
| 53rdof 62 | o4-mini | mean_score (high) | 4.878 |
| 54thof 62 | Claude Opus 4.5 | mean_score | 4.878 |
| 55thof 62 | GPT-5.5 Instant | mean_score | 2.439 |
| 56thof 62 | DeepSeek V4 Pro | mean_score (max) | 2.439 |
| 57thof 62 | GPT-5 nano | mean_score (high) | 2.439 |
| 58thof 62 | Claude Opus 4.1 | mean_score | 2.439 |
| 59thof 62 | Claude Sonnet 4.5 | mean_score | 2.439 |
| 60thof 62 | Gemini 3.5 Flash-Lite | mean_score (high) | 0 |
| 61stof 62 | Gemini 2.5 Pro | mean_score | 0 |
| 62ndof 62 | o3-mini | mean_score (high) | 0 |
Read from the board on 2026-09-16