← LLM ModelsOpen SourceView on HF →→
Phi 4
Microsoft
[Microsoft Research](/microsoft) Phi-4 is designed to perform well in complex reasoning tasks and can operate efficiently in situations with limited memory or where quick responses are needed. At 14 billion...
huggingface.co ↗📖Context16K tokens
🚀Speed-
💵Input price$0.065/1M
💸Output price$0.140/1M
🧠Parameters14.66B
Benchmarks
Gray bar = dataset averageComposite ScoreScore = MMLU×20% + GPQA×30% + HumanEval×25% + SWE-Bench×25%-
HF Open LLM Leaderboard
IFEval · BBH · MATH · GPQA · MuSR · MMLU-Pro
Average
40.7
IFEvalInstruction following
68.8%avg 79.9
BBHBig Bench Hard
55.3%avg 48.7
MATH Lvl 5Competition mathematics
50.0%avg 37.8
GPQAGraduate-level science Q&A
11.5%avg 12.9
MuSRMultistep soft reasoning
10.1%avg 10.7
MMLU-ProProfessional knowledge
48.6%avg 40.0
Made by
🏢
Microsoft
startupMore from Microsoft
API Providers
Details
- Provider
- Microsoft
- Parameters
- 14.66B
- Context
- 16K tokens
- Speed
- -
- Open Source
- Yes
- License
- Open (varies)
- Released
- Jan 2025

