Phi-4 is an open-weights model from Microsoft, released December 12, 2024. It has an ECI score of 130 and ranks 149th of 249 tracked models. At the time of its release, it ranked 7th of 100 tracked models. At $0.07/Mtok for input tokens and $0.14/Mtok for output tokens, it is less expensive per token than similarly-capable models.
| Domain | Benchmark | ||||
|---|---|---|---|---|---|
ECI scoreBest score 166 | 130 | 122 | 130 | 129 | |
| Ranking | #149 | #185 | #148 | #157 | |
| Mathematics | MATH Level 5Best score 98.1% | ||||
OTIS Mock AIME 2024-2025Best score 100.0% | |||||
| Science | GPQA DiamondBest score 95.8% | ||||
| World knowledge | MMLUBest score 88.1% | ||||
| Writing & creativity | Lech Mazur WritingBest score 8.6 | 6.3 | — | 6.4 | 8.2 |
| Other | LiveBenchBest score 82.3% |