Phi-4

14B model that punches well above its weight

Phi-4 is Microsoft's 14-billion parameter model that achieves remarkable benchmark scores, especially on STEM and reasoning tasks, by training on carefully curated synthetic data rather than raw web text. It outperforms models 5× its size on mathematics and coding.

Try Phi-4

Specifications

Context window16k tokens
Input price$0.070 / 1M tokens
Output price$0.140 / 1M tokens
Credits per query2 cr
Released2024-12
LicenseMIT

Features

Tool useJSON modeFunction callingCode

Benchmark Scores

What do these mean?

Sourced from official model cards and academic papers. Higher is better.

PhD-level science & reasoning

Mathematical olympiad problems

Chatbot Arena human preference ranking

Python coding — pass@1 rate

82.6%

Competition-level mathematics

80.4%

General knowledge across 57 academic domains

84.8%

Multi-turn instruction following (out of 10)

Real-world GitHub issue resolution rate

Run locally

Download and run this model on your own hardware — no API key or internet required after setup.

Min VRAM8 GB
Rec. VRAM10 GB
Best VRAM15 GB
Parameters14B
Speed~22 tok/s
Context16K tokens

Ollama

Terminal

Run this command in your terminal. Ollama downloads and manages the model locally.

ollama pull phi4