Skip to content
AI Model Radar

Phi 4 mini reasoning

microsoft/Phi-4-mini-reasoning
Open Source · chat · open-weights
GA Alert me on changes
Intelligence
3
Context
32.8K
Max output
—
Pricing
Open weights
Modalities
text
Released
29 Apr 2025
Intelligence Index via Artificial Analysis · 0–100, higher is better
License: mit · microsoft/Phi-4-mini-reasoning
AI summary
● machine-written

Phi-4-mini-reasoning: 3.8B parameter open-weights model for math and reasoning

Phi-4-mini-reasoning is a lightweight open-source model from Microsoft designed for reasoning-dense tasks in mathematics, science, and code. It uses supervised fine-tuning on chain-of-thought traces and reinforcement learning to optimize for structured reasoning problems. The model is released under an MIT license and available as open weights.

What's new
  • Combines supervised fine-tuning on chain-of-thought traces with reinforcement learning
  • Optimized for math, science, and code reasoning tasks
  • 3.8 billion parameters with efficient inference
  • Released April 29, 2025
Best for
Math problem solvingScience reasoning tasksCode generation and analysisChain-of-thought reasoning
Sources

Source: https://huggingface.co/microsoft/Phi-4-mini-reasoning