Benchmark Scores
82.2/100MMLU84.8
HumanEval82.6
Math80.4
Reasoning81
Coding82
About Phi-4
Phi-4 is Microsoft's 14B parameter model that punches far above its weight, particularly on math and reasoning benchmarks. It demonstrates that data quality can compensate for model size, excelling on MATH and competition benchmarks.
Strengths
- +Exceptional for its size
- +Strong math for 14B model
- +Efficient local deployment
- +MIT license
Weaknesses
- -Small context window
- -Limited on complex tasks
- -Narrow training data focus
Pricing
Per 1M tokens
Input$0.07
Output$0.14
Quick Facts
- Context Window
- 16K
- Parameters
- 14B
- Release Date
- 2024-12-12
- Category
- Open Source
Compare Models
See how Phi-4 stacks up against other models