Vpct — Epoch AI
Vpct — Epoch AI is run by Epoch AI. It has ranked 38 models, of which the catalogue holds 24, scoring from 0.3 to 0.91.
measured by Epoch AI · the board itself
| Place | Model | Metric | Score |
|---|---|---|---|
| 1stof 38 | gemini-3-pro-preview | Correct | 0.91 |
| 2ndof 38 | GPT-5.2 | Correct (xhigh) | 0.84 |
| 3rdof 38 | Gemini 3 Flash Preview | Correct | 0.726 |
| 5thof 38 | GPT-5 | Correct (high) | 0.66 |
| 7thof 38 | GPT-5.1 | Correct (high) | 0.587 |
| 8thof 38 | o4-mini | Correct (medium) | 0.575 |
| 10thof 38 | o3 | Correct (medium) | 0.52 |
| 11thof 38 | Gemini 2.5 Pro | Correct | 0.48 |
| 12thof 38 | Gemini 2.5 Pro Preview 06-05 | Correct | 0.464 |
| 16thof 38 | Gemini 2.5 Pro Preview 05-06 | Correct | 0.405 |
| 17thof 38 | GPT-5 mini | Correct (medium) | 0.402 |
| 18thof 38 | GPT-4o (2024-11-20) | Correct | 0.4 |
| 19thof 38 | Claude Opus 4.5 | Correct | 0.4 |
| 20thof 38 | Claude Sonnet 4.5 | Correct | 0.398 |
| 21stof 38 | Claude 3.7 Sonnet | Correct | 0.39 |
| 23rdof 38 | Gemini 2.5 Flash | Correct | 0.38 |
| 24thof 38 | Claude Opus 4 | Correct | 0.38 |
| 26thof 38 | GPT-5 nano | Correct (high) | 0.372 |
| 27thof 38 | o1 | Correct (medium) | 0.37 |
| 30thof 38 | Claude Opus 4.1 | Correct | 0.35 |
| 31stof 38 | GPT-4o-mini (2024-07-18) | Correct | 0.34 |
| 32ndof 38 | Claude Sonnet 4 | Correct | 0.34 |
| 33rdof 38 | Claude 3.5 Sonnet | Correct | 0.33 |
| 38thof 38 | Gemini 2.5 Flash-Lite Preview | Correct | 0.3 |
Read from the board on 2026-09-16