Home/Models/phi-4
LLMOpen SourceOpen SourceSLMReasoning

Phi-4

by Microsoft · 14B model that punches above its weight.

Visit website ↗
RAI ScoreA+Excellent
860/ 1000
↑ 72RAI7dMethodology
Composite of 8 signals · 2,481 contributors

// Live community signal

Rate this model

4.3
RAI · live
Sign in to cast your signal →

About

Microsoft's 14B reasoning-tuned SLM, competitive with much larger frontier models on math.

Dimension Scores

intelligence96
creativity88
accuracy94
speed72
coding98
research92
value84
safety90

Pros

  • +Excellent reasoning for size
  • +Open weights
  • +Runs on consumer GPUs

Cons

  • Short context
  • Weak multilingual

Community Reviews

Reviews open once this model is indexed in the live database.

Similar AIs