Individual benchmark scores plotted by date.
| Organisation | Model | Reported | Top Score | Info | Self Reported | Source |
|---|---|---|---|---|---|---|
| DeepSeek V4 Pro | 24 Apr 2026 | 90.20% | Best reported mode. Non-think: 9.2, High: 85.5, Max: 90.2 | Yes | Source | |
| DeepSeek V4 Flash | 24 Apr 2026 | 85.70% | Best reported mode. Non-think: 9.3, High: 72.1, Max: 85.7 | Yes | Source |
