Microsoft: Phi 4
[Microsoft Research](/microsoft) Phi-4 is designed to perform well in complex reasoning tasks and can operate efficiently in situations with limited memory or where quick responses are needed. At 14 billion...
Everything else
| Max output | 16K |
|---|---|
| Input / 1M | $0.07 |
| Output / 1M | $0.14 |
| Cache read / 1M | — |
| Cache write / 1M | — |
| 1M in + 1M out | $0.21 |
| Takes | text |
| Returns | text |
| Tools | no |
| Reasoning | no |
| Open weights | yes |
| Released | 2025-01-10 |
| Knowledge cutoff | 2024-06-30 |
| Retires | — |
| Catalog id | microsoft/phi-4 |
Benchmarks
| source | epoch |
|---|---|
| eci | 131.07 |
| confidence | Confident |
| accessibility | Open weights (unrestricted) |
| gpqa | 0.561 |
| gpqaStderr | 0.0259 |
| mathLevel5 | 0.649 |
| mathLevel5Stderr | 0.0107 |
| aime | 0.138 |
| aimeStderr | 0.037 |
Benchmark scores by Epoch AI, CC BY 4.0.
Using it
In 00, this model is picked per agent — and per tier, so the model that answers your customers need not be the one that writes your code. Get 00.