| Benchmark | GPT-4o mini | GPT-5.4 |
|---|---|---|
| AA Agentic Index | 1 | 58.2 |
| AA Intelligence | 7 | 53.1 |
| ARC-AGI-2 | 0 | 74 |
| Artificial Analysis Coding Index | 11.4 | 71.1 |
| gdpval | 0 | 50.1 |
| GPQA Diamond | 42.6 | 93 |
| HLE | 4.2 | 43.7 |
| IFBench | 31 | 73.9 |
| MMMU-Pro | 41.5 | 81.2 |
| scicode | 22.9 | 56.6 |
| TauBench V3 - Banking | 2.9 | 39.6 |
| Terminal-Bench 2.1 | 5.6 | 78.3 |
| τ³-Bench | 2.9 | 72.9 |
Best tracked score per model per benchmark (default configuration; source-attributed). ↓ marks lower-is-better metrics. Open either model for its full surface, provenance and pricing.