Claude Opus 4 vs DeepSeek-R1 0528
Wins 3 of 4 areas
Coding · Long documents · Following instructions
Wins 1 of 4 areas
Reasoning
Claude Opus 4 is the stronger all-rounder.DeepSeek-R1 0528 is cheaper and better at reasoning.
Scores updated · 39 tests both models report · How we compare
Where each one wins
Tests won in each of the four areas where both have results. Each piece is one test, so a longer bar means more evidence; grey means the two scored within a point of each other.
- CodingWriting and fixing software20Claude Opus 42 of 3 tests · 1 tie
- Long documentsFinding answers in very long texts20Claude Opus 42 of 2 tests
- Following instructionsDoing exactly what it is asked20Claude Opus 42 of 2 tests
- ReasoningHard problems that need careful thinking12DeepSeek-R1 05282 of 3 tests
What it costs
Prices per million tokens, roughly 750,000 words. The bars show the cost of a million tokens read plus a million written.
DeepSeek-R1 0528 costs 95% less for the same work.
The biggest differences
The tests each model wins by the widest margin, up to three each. Scores are out of 100.
Where Claude Opus 4 pulls ahead
- Fixes real GitHub issues in Python projectsSWE-bench Verified+27.9points ahead
- Common-sense trick questionsSimpleBench+18points ahead
- Hard command-line tasks in a real terminalTerminal-Bench Hard+15.2points ahead
Where DeepSeek-R1 0528 pulls ahead
- Very hard expert questions across many subjectsHumanity's Last Exam+3.5points ahead
- Graduate-level biology, physics and chemistry questionsGPQA Diamond+1.7points ahead
Every test, side by side
All 39 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingClaude Opus 4
- SWE-bench VerifiedClaude Opus 4 by 27.972.544.6+27.9
- Terminal-Bench HardClaude Opus 4 by 15.231.115.9+15.2
- SciCodetie39.840.3tie
ReasoningDeepSeek-R1 0528
- SimpleBenchClaude Opus 4 by 1858.840.8+18
- Humanity's Last ExamDeepSeek-R1 0528 by 3.512.315.8+3.5
- GPQA DiamondDeepSeek-R1 0528 by 1.779.681.3+1.7
Long documentsClaude Opus 4
- AA-LCRClaude Opus 4 by 13.669.355.7+13.6
- LongBench v2Claude Opus 4 by 3.555.652.1+3.5
Following instructionsClaude Opus 4
- IFBenchClaude Opus 4 by 14.153.739.6+14.1
- Multi-ChallengeClaude Opus 4 by 13.658.645+13.6
Other results29 tests, not counted
Tests outside the eight areas. They are not counted above: several are summary scores built from other tests, or the same test under another name.
- SimpleQADeepSeek-R1 0528 by 69.522.892.3+69.5
- HMMT 2025DeepSeek-R1 0528 by 63.515.979.4+63.5
- AIME 2024DeepSeek-R1 0528 by 43.248.291.4+43.2
- Terminal-BenchClaude Opus 4 by 37.543.25.7+37.5
- HarmBenchClaude Opus 4 by 37.191.754.6+37.1
- τ²-Bench Telecom (AA run)Claude Opus 4 by 36.973.436.5+36.9
- ZebraLogicDeepSeek-R1 0528 by 35.859.395.1+35.8
- livecodebench_hardDeepSeek-R1 0528 by 30.533.163.6+30.5
- TAU-bench (retail)Claude Opus 4 by 17.581.463.9+17.5
- HMMT Feb. 2025DeepSeek-R1 0528 by 16.76076.7+16.7
- LiveCodeBench (24/8~25/5)DeepSeek-R1 0528 by 16.556.673.1+16.5
- ARC-AGI-1Claude Opus 4 by 14.535.721.2+14.5
- AIME 2025DeepSeek-R1 0528 by 1275.587.5+12
- livecodebench_mediumDeepSeek-R1 0528 by 10.580.490.9+10.5
- Artificial Analysis Coding IndexClaude Opus 4 by 103424+10
- AA IntelligenceClaude Opus 4 by 7.520.613.1+7.5
- TAU-bench (airline)Claude Opus 4 by 6.159.653.5+6.1
- MATH-500 (EM)DeepSeek-R1 0528 by 3.694.498+3.6
- LiveCodeBenchDeepSeek-R1 0528 by 2.970.473.3+2.9
- OpenAI-MRCR (128k)DeepSeek-R1 0528 by 2.648.951.5+2.6
- XSTestDeepSeek-R1 0528 by 1.89798.8+1.8
- simple_safety_testsClaude Opus 4 by 1.710098.3+1.7
- MMLU-ProClaude Opus 4 by 1.686.685+1.6
- Aider-Polyglottie70.771.6tie
- FullStackBenchtie70.369.4tie
- MMLU-Reduxtie94.293.4tie
- bbqtie97.196.5tie
- anthropic_red_teamtie98.999.1tie
- livecodebench_easytie99.199.2tie
Questions people ask
Which is better, Claude Opus 4 or DeepSeek-R1 0528?
Claude Opus 4 wins three of the four areas where both have results: coding, long documents and following instructions. DeepSeek-R1 0528 wins reasoning, and costs 95% less.
Which is better for coding?
Claude Opus 4. It wins 2 of the 3 coding tests both models report; DeepSeek-R1 0528 wins none, and 1 is a tie.
Which is cheaper?
Claude Opus 4 costs $15.00 per million input tokens and $75.00 per million output tokens; DeepSeek-R1 0528 costs $1.35 and $3.00. That makes DeepSeek-R1 0528 about 95% cheaper for the same work.
How do you compare the two?
We use the 39 benchmark tests both models have published scores on. The verdict counts the 10 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 29 are listed but not counted, because several are summary scores or repeat a test. Each score is the one shown on the model's own page.