DeepSeek V3

deepseek · Ranked across 6 benchmarks · best rank #4

Benchmark scores

BenchmarkCategoryRankScoreCaptured
OpenRouter · Weekly Usage usage #4 #5 2026-05-02
Aider Polyglot code #5 74.2% 2025-10-03
Chatbot Arena · Open Weights chat #7 1424 2026-04-30
Terminal-Bench 2.0 agents #12 39.6% 2026-02-08
SWE-bench Verified agents #13 70.0% 2026-02-17
Chatbot Arena chat #20 1424 2026-04-30