Up to four models
Claude Opus 4 vs Claude 3 Opus
On the index (GPT-4 = 100), Claude Opus 4 scores 136 and Claude 3 Opus 108. Side by side on the same third-party tests, with price and context. The link in your address bar shares this exact comparison.
Claude Opus 4 leads Claude 3 Opus by 28 points, ahead on 2 of 2 shared core tests.
Test by testthird-party results
Claude Opus 4Claude 3 Opus
Every result as a table
| Test | Claude Opus 4 | Claude 3 Opus |
|---|---|---|
| Reasoning & knowledge | ||
| GPQA Diamond (Epoch AI) | 76.3% | 47.2% |
| SimpleBench | 58.8% | 23.5% |
| Maths | ||
| AIME-style maths (OTIS mock) | 64.4% | 4.7% |
| MATH (level 5) | 85.0% | 37.5% |
| Coding | ||
| WeirdML | 43.7% | 19.2% |
| Agents | ||
| DTBench | 81.6% | 61.6% |
| LMCA | 37.4% | 17.0% |
Task length · METR38 sec18 sec
Tests at least two of these models have taken. Bold rows are the core tests.