DeepSeek R1 0528 — Benchmarks

Benchmark scores for DeepSeek R1 0528 aggregated from public leaderboards, with how it ranks among open models. See hardware requirements for what you need to run it.

Overall rank: #11 of 123 open modelscomposite 69.5/100 across 7 benchmarks in 4 categories · methodology

Coding

BenchmarkScoreOpen rankAll models
Aider Polyglot71.4%#2 / 18#19 / 69

Knowledge

BenchmarkScoreOpen rankAll models
MMLU-Pro83.4%#11 / 119#46 / 259

Math

BenchmarkScoreOpen rankAll models
AIME 2024/202566.4%#28 / 74#137 / 270
MATH Level 596.6%#1 / 32#9 / 108

Reasoning

BenchmarkScoreOpen rankAll models
GPQA Diamond76.3%#23 / 83#132 / 291
ARC-AGI21.2%#11 / 16#163 / 200
SimpleBench40.8%#13 / 23#70 / 101

Scores aggregated from public benchmark sources (each linked from the benchmark pages). llmrun does not run these benchmarks.