Local LLM model page

Phi-4 Mini (3.8B)

Microsoft's latest small miracle. Punches way above its weight in reasoning & code.

Find the best model for my hardwareBrowse all 183 LLMs
Parameters
3.8B
Minimum RAM
4 GB
Model size
2.5 GB
Quantization
Q5_K_M

Can Phi-4 Mini (3.8B) run locally?

Phi-4 Mini (3.8B) is best suited for entry-level laptops and desktops. LocalClaw recommends Q5_K_M as the default quantization, with at least 4 GB RAM.

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

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

chatcodelightspeed

Strengths

  • MIT license
  • Punches way above weight class
  • 128K context at 3.8B
  • Great at reasoning & code for its size

Limitations

  • English-only
  • Limited factual knowledge
  • Can hallucinate

Best use cases

  • Quick coding help
  • Reasoning on-the-go
  • Edge deployment
  • Education

Benchmarks

Speed: 9/10

Quality: 6/10

Coding: 7/10

Reasoning: 6/10

Technical details

Developer: Microsoft Research

License: MIT

Context window: 131,072 tokens

Architecture: Transformer decoder-only, 128K context

Released: 2025-02

Similar models

qwen3-4bgemma3-4bllama3.2-3b