BenchClub
OverviewLeaderboardModelsLabsCreatorsTest sets
Sign in
P

phi-2

Microsoft
Hugging FaceArtificial AnalysisCompare

LMArena

Not ranked on LMArena

LMArena only ranks the models it serves.

Built on it

Quantizations
64
Fine-tunes
416
Adapters
994
Merges
8

Ways to run it

  • GGUF · llama.cpp, Ollama, LM StudioTheBloke/phi-2-GGUF32 builds
  • MLX · Apple silicon: mlx-lm, LM Studionexsendev/webicoder-v3-mlx-q86 builds
  • GPTQ · vLLM, SGLangTheBloke/phi-2-GPTQ3 builds

Most downloaded

  • phi-2-GGUFTheBloke · Quantization18.8K downloads
  • phi-2-llamafilejartine · Fine-tune2.5K downloads
  • phi-2-michuggingface-mic · Fine-tune2.3K downloads
  • phi-2-q4f16_1-MLCmlc-ai · Quantization1.8K downloads
  • phi-2-GGUFsecond-state · Quantization792 downloads

BenchClub runs

Not runnable yet

Runs open once a phi-2 build is registered.

More from Microsoft

See more
  • PPhi-3.5-mini-instruct324.4K3.8B
  • PPhi-3-mini-4k-instruct314.8K3.8B
  • PPhi-3-mini-128k-instruct241.1K3.8B
  • Pphi-4205.7K14.7B

Other ≤3B models

See more
  • MMiniCPM5-2BOpenBMB986.8K2.5B
  • LLFM3-1.7BLiquid AI110K1.7B
  • SSpark-X2.6-1.7BXHToken5.7K1.7B
  • MMNN-Pocket-0.3Btaobao-mnn5.4K362M

Details

Parameters
2.8B
Size class
≤3B
Capabilities
Text
License
MIT
Access
Open
Downloads, last 30 days
177,357
Likes
3,537
Released
Dec 13, 2023
Last updated
Dec 8, 2025

Usage

Hugging Face, as of Nov 18, 2026

Downloads, all time
23,980,997
Trending score
2
BenchClub

Ran it, or it didn’t happen.

Explore

  • Overview
  • Leaderboard
  • Cloud leaderboard
  • Models
  • Labs
  • Releases
  • What fits
  • Families
  • Licenses
  • Most built on
  • Formats
  • API usage
  • Compare
  • Discussions
  • Creators
  • Test sets

Take part

  • Run a benchmark
  • Methodology
  • Sign in

Data sources

  • Hugging FaceModel metadata
  • LMArenaArena scores, CC BY 4.0
  • OpenRouterModel usage, CC BY 4.0
  • How we use them

© 2026 BenchClub