Skip to content
Model Wayfinder
← Back to comparison
Microsoft

Phi 4

Open weightsMIT · commercial use allowed

Benchmarks

Benchmark average
4/100
based on 5 benchmarks
BenchmarkScorePositionSource
LMArena Text conversation (user votes)1,217 Elo3/ 150LMArena · Nov 11, 2026
LMArena Creative Writing creative writing1,181 Elo3/ 150LMArena · Nov 11, 2026
LMArena Non-English languages1,197 Elo3/ 150LMArena · Nov 11, 2026
LMArena Coding coding1,231 Elo6/ 150LMArena · Nov 11, 2026
LMArena Math maths1,246 Elo6/ 145LMArena · Nov 11, 2026

Position: 100 = best, 0 = last among all models listed here that were tested on this benchmark.

Strengths & weaknesses

Automatic summary from benchmark positions – not an opinion.

Pros
    Cons
    • −Weaker in coding (position 6/100)
    • −Weaker in maths (position 6/100)
    • −Weaker in conversation (user votes) (position 3/100)
    • −Weaker in languages (position 3/100)
    • −Weaker in creative writing (position 3/100)

    Our quality rating

    Chatbot (e.g. customer service)1/3
    Writing texts and emails1/3
    Summarising documents1/3
    Translating1/3
    Programming1/3
    Maths & calculations1/3
    Field: Law & public administration1/3
    Field: Medicine & health1/3
    Field: Business & finance1/3
    Field: Science & research1/3

    Tiers derived from: LMArena Text 2026-11-11 · LMArena Creative Writing 2026-11-11 · LMArena Longer Query 2026-11-11 · LMArena Non-English 2026-11-11 · LMArena Coding 2026-11-11 · LMArena Math 2026-11-11 · LMArena Legal & Government 2026-11-11 · LMArena Medicine & Healthcare 2026-11-11 · LMArena Business & Finance 2026-11-11 · LMArena Science 2026-11-11

    Source, checked Nov 17, 2026