AI Hub
All models
LLMOpen weights

Phi-2

Microsoft

Updated May 21, 2026 · Benchmarks via Artificial Analysis, specs & pricing via OpenRouter · Methodology · Report an error

Superseded by Phi-3

Specifications

Type
LLM
Access
Open weights
Released
December 12, 2023
Parameters
2.7B
Context window
2,048 tokens
Input
Text
Output
Text

Our take

A 2.7B model that punched far above its size by training on carefully curated, textbook-quality data — evidence that data quality can rival scale.

Compare Phi-2 with

See all Phi-2 alternatives →