AI model · by Microsoft
Phi 4
Phi 4 is Microsoft's language model meant to advance language model research and serve as a foundation for generative AI features.
- Released 12 Dec 2024
- 16K tokens context
- Open weights · MIT
- Checked against Microsoft's page 19 Oct 2026
What Phi 4 is for
Microsoft says it built Phi 4 to speed up language model research. The company intends it as a foundation for AI-powered generative features. Microsoft says it is not meant to support multilingual use.
What Microsoft says it does well
- Underwent supervised fine-tuning and direct preference optimization for precise instruction adherence and robust safety measures.
Limits Microsoft mentions
- Context length of 16K tokens.