Microsoft model product
Phi 4 Mini Instruct is a lightweight (3.8B parameters) open model built upon synthetic data and filtered web data, focusing on high-quality reasoning. It supports a 128K token context length and is enhanced for instruction adherence and safety via supervised fine-tuning and direct preference optimization.
Updated Aug 10, 2026. Default version: Phi 4 Mini
Structured fields from the published default version.
This profile uses the latest version under this unique model that has a calculated LLMBoard score. Arena and price are excluded.
Published benchmark records for the scored version Phi 4 Mini.
| OpenBookQA | 0.8 | 3 | 5 | 50.0% | C | |
| Social IQa | 0.7 | 3 | 9 | 75.0% | C | |
| TruthfulQA | 0.7 | 4 | 18 | 82.3% | C | |
| Multilingual MMLU | 0.5 | 5 | 5 | 0.0% | C | |
| BoolQ | 0.8 | 7 | 10 | 33.3% | C | |
| PIQA | 0.8 | 10 | 11 | 10.0% | C | |
| BIG-Bench Hard | 0.7 | 12 | 21 | 45.0% | C | |
| ARC-C | 0.8 | 15 | 34 | 57.6% | C | |
| Winogrande | 0.7 | 19 | 22 | 14.3% | C | |
| Arena Hard | 0.3 | 24 | 26 | 8.0% | C | |
| MGSM | 0.6 | 24 | 31 | 23.3% | C | |
| HellaSwag | 0.7 | 26 | 27 | 3.9% | C | |
| GSM8k | 0.9 | 32 | 48 | 34.0% | C | |
| MATH | 0.6 | 45 | 71 | 37.1% | C | |
| MMLU | 0.7 | 90 | 100 | 10.1% | C | |
| MMLU-Pro | 0.5 | 111 | 129 | 14.1% | C | |
| GPQA | 0.3 | 227 | 233 | 2.6% | C |
Preference and agent-evaluation signals from published Arena datasets.
The default version has no published Arena rows, or its source alias has not been resolved.
Official vendor API PAYG pricing is summarized first. The table then lists individual provider offerings without treating their minimum as the official price.
| Azure Cognitive Services | phi-4-mini | global | $0.075 | $0.30 | 128K | |
| Azure | phi-4-mini | global | $0.075 | $0.30 | 128K |
Official prices use only the vendor's configured official Provider and positive standard USD PAYG rates. Third-party offers remain explicitly labeled.
All published versions linked to this unique model. The score columns identify the version used by the current overall ranking.
| Phi 4 Mini | 31.5 | 3.8B | 128K | 4.1K | Yes | MIT |
A concise description based on the published model registry.
Phi 4 Mini Instruct is a lightweight (3.8B parameters) open model built upon synthetic data and filtered web data, focusing on high-quality reasoning. It supports a 128K token context length and is enhanced for instruction adherence and safety via supervised fine-tuning and direct preference optimization.
Use the benchmark, Arena and pricing sections above as separate evidence. A missing field means the current data snapshot does not support that claim.
Data snapshot: 2026-08-07. Editorial model content is not available in the backend.
Open a comparison with the three ranked models immediately above and below this model.
Recommendations prioritize the same model type and family, then the closest published LLMBoard score.
Common questions about Phi 4 Mini.
Phi 4 Mini's default version was released on Feb 1, 2025.
Phi 4 Mini's official API price is $0.075 per million input tokens and $0.30 per million output tokens via Azure. The lowest tracked third-party offer starts at $0.075 input and $0.30 output via Azure Cognitive Services.
Phi 4 Mini is published under Microsoft in the model registry.
The default version has a 128K token context window.
Yes. The default version is marked as open weight under MIT.
2 published provider offerings are linked to the default version.
Nearby ranked alternatives include QwQ 32B, Qwen2.5 7B, Llama 4 Scout.