Local LLM model page

Phi-3 (3.8B)

Microsoft lightweight powerhouse. Punches way above its weight. 11.3M downloads. Great for edge devices.

Parameters
3.8B
Minimum RAM
6 GB
Model size
2.3 GB
Quantization
Q5_K_M

Can Phi-3 (3.8B) run locally?

Phi-3 (3.8B) is best suited for entry-level laptops and desktops. LocalClaw recommends Q5_K_M as the default quantization, with at least 6 GB RAM.

Search term for LM Studio or compatible runtimes: phi-3-mini-4k-instruct

Hugging Face repository: lmstudio-community/Phi-3-mini-4k-instruct-GGUF

chatreasoninglightspeed

Strengths

  • 11.3M downloads
  • MIT license
  • Great for edge devices
  • Pioneered small-model-big-performance

Limitations

  • Only 4K context
  • English-only
  • Superseded by Phi-4

Best use cases

  • Quick Q&A
  • Edge deployment
  • Education
  • Lightweight tasks

Benchmarks

Speed: 9/10

Quality: 6/10

Coding: 6/10

Reasoning: 6/10

Technical details

Developer: Microsoft Research

License: MIT

Context window: 4,096 tokens

Architecture: Transformer decoder-only

Released: 2024-04