Skip to content
Model Wayfinder
← Back to comparison
DeepSeek

DeepSeek V3.2

Open weightsMIT · commercial use allowed

Benchmarks

Benchmark average
47/100
based on 6 benchmarks
BenchmarkScorePositionSource
LMArena Text conversation (user votes)1,425 Elo52/ 150LMArena · Nov 11, 2026
LMArena Creative Writing creative writing1,400 Elo53/ 150LMArena · Nov 11, 2026
LMArena Non-English languages1,409 Elo51/ 150LMArena · Nov 11, 2026
LMArena Coding coding1,453 Elo53/ 150LMArena · Nov 11, 2026
LMArena Math maths1,435 Elo53/ 145LMArena · Nov 11, 2026
LMArena WebDev coding1,362 Elo17/ 108LMArena · Nov 15, 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

      No clear strengths or weaknesses – solid across the board.

      Our quality rating

      Chatbot (e.g. customer service)3/3
      Writing texts and emails3/3
      Summarising documents3/3
      Translating3/3
      Programming3/3
      Maths & calculations3/3
      Field: Law & public administration3/3
      Field: Medicine & health3/3
      Field: Business & finance3/3
      Field: Science & research3/3

      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