The Model Index
Up to four models

DeepSeek-R1-0528 vs DeepSeek-R1

On the index (GPT-4 = 100), DeepSeek-R1-0528 scores 135 and DeepSeek-R1 130. Side by side on the same third-party tests, with price and context. The link in your address bar shares this exact comparison.

Next to DeepSeek-R1-0528
Presets
DeepSeek-R1-0528
135120–151
WeightsOpen
ReleasedMay 2025
Price in / out–
Context–
Core tests taken2 of 9
DeepSeek-R1
130114–146
WeightsOpen
ReleasedJan 2025
Price in / out–
Context–
Core tests taken2 of 9

DeepSeek-R1-0528 leads DeepSeek-R1 by 6 points, ahead on 2 of 2 shared core tests.

Test by testthird-party results

DeepSeek-R1-0528DeepSeek-R1
0%25%50%75%100%REASONING & KNOWLEDGEGPQA Diamond (Epoch AI)GPQA Diamond (Epoch AI)Fiction.LiveBenchFiction.LiveBenchSimpleBenchSimpleBenchMATHSAIME-style maths (OTIS mock)AIME-style maths (OTIS mock)MATH (level 5)MATH (level 5)CODINGAider PolyglotAider PolyglotWeirdMLWeirdML
Every result as a table
TestDeepSeek-R1-0528DeepSeek-R1
Reasoning & knowledge
GPQA Diamond (Epoch AI)76.3%71.7%
Fiction.LiveBench75.0%69.4%
SimpleBench40.8%30.9%
Maths
AIME-style maths (OTIS mock)66.4%53.3%
MATH (level 5)96.6%93.0%
Coding
Aider Polyglot71.4%56.9%
WeirdML41.6%36.5%
Task length · METR32 sec31 sec

Tests at least two of these models have taken. Bold rows are the core tests.

This page as a chartDownload PNG