Phi-4 Specs, Benchmarks & Downloads
Phi-4
Microsoft reasoning model
Status
active
Parameters
14.00
Context Window
16384
Modality
Text
Release Date
Dec 12, 2024
About
Explore Phi-4, a text AI model developed by Microsoft. View specifications, context window, architecture, license, supported platforms, features, benchmarks, official downloads, and related articles.
Phi-4 improves mathematical reasoning and complex problem solving.
Family
PhiDeveloper
MicrosoftArchitecture
TransformerLicense
Apache 2.0Hardware Requirements
Minimum:
Use Case: Local inference
GPU Minimum: RTX 3060
GPU VRAM: 12GB
System RAM: 32GB
Storage: 40GB
Operating System: Linux
Phi-4 deployment
Recommended:
Use Case: Desktop inference
GPU Minimum: RTX 4090
GPU VRAM: 24GB
System RAM: 64GB
Storage: 60GB
Operating System: Linux
Recommended local deployment
High Performance:
Use Case: Production inference
GPU Minimum: RTX 4090
GPU VRAM: 24GB
System RAM: 128GB
Storage: 100GB
Operating System: Linux
High performance setup
Model Deployment Readiness
Model Ecosystem Support
15%
Platforms
25%
Features
1%
Downloads
20%
Platform Compatibility
25%
Supported across 3 of 12 platforms
Context Window
Industry-standard context window comparison
16,384
tokens
16K
4K
32K
128K
1M
2M
Database Rank
#44
Percentile
Top 57%
Capacity
Medium
Benchmark Performance
Evaluation profile across publicly reported benchmarks
|
MMLU
|
Knowledge | 84.8 % | 2024 |
|
GPQA
|
Reasoning | 56.1 % | 2024 |
|
HumanEval
|
Coding | 82.6 % | 2024 |
|
MATH-500
|
Mathematics | 80.4 % | 2024 |
|
GSM8K
|
Mathematics | 91.4 % | 2024 |
|
IFEval
|
Instruction Following | 88.4 % | 2024 |
|
BBH
|
Reasoning | 79.3 % | 2024 |
|
GPQA
|
Reasoning | 56.1 % | 2024 |
|
MATH-500
|
Mathematics | 80.4 % | 2024 |
|
GSM8K
|
Mathematics | 80.4 % | 2024 |
Highest Benchmark
GSM8K
91.4%
Benchmarks Available
10
Latest Evaluation
Dec 2024
