Updated daily at 05:00 UTC
- Released
- 20 Aug 2024
- Status
- Legacy
- Context
- 131.1K
- Max output
- —
- Input per 1M tokens
- —
- Output per 1M tokens
- —
- Open weights
- Yes
- License
- mit
- Parameters
- 4.15B
- Knowledge cutoff
- —
- Reasoning
- No
- Modalities
- text, image → text
Category scores
Out of 100, from independent and third-party results only, by method category-2026-09-29.1.
| Category | Score | Rank | Coverage |
|---|---|---|---|
| Human preference | 5.9 | Not ranked | 1 of 5 |
Benchmark results
Every result we hold for this model, with who ran it and where it was published.
Knowledge & reasoning
| Benchmark | Metric | Value | Provenance | Setting | Evaluated | Source |
|---|---|---|---|---|---|---|
| ScienceQA | Accuracy | 91.3% | Self-reported | — | — | Phi-3 Technical Report: A Highly Capable Language Model Locally on Your Phone via Epoch AI |
Human preference
| Benchmark | Metric | Value | Provenance | Setting | Evaluated | Source |
|---|---|---|---|---|---|---|
| Vision Arena (arena.ai) | Arena score (Elo scale) | 920 | Independent | — | 28 Sep 2026 | Arena |
| Vision Arena (arena.ai) | Votes | 2592 | Independent | — | 28 Sep 2026 | Arena |
Other Phi Vision / Multimodal versions
- Phi-4-reasoning-vision-15BJan 2026
- Phi-4-multimodal26 Feb 2025
- Phi-3-vision21 May 2024