Claude Sonnet 4 vs DeepSeek-R1 0528
Wins 4 of 6 areas
Coding · Agents · Long documents · Following instructions
Wins 1 of 6 areas
Reasoning
Claude Sonnet 4 is the stronger all-rounder.DeepSeek-R1 0528 is cheaper and better at reasoning.
Scores updated · 43 tests both models report · How we compare
Where each one wins
Tests won in each of the six 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 software30Claude Sonnet 43 of 4 tests · 1 tie
- Following instructionsDoing exactly what it is asked20Claude Sonnet 42 of 2 tests
- AgentsCarrying out multi-step tasks on its own10Claude Sonnet 41 of 2 tests · 1 tie
- Long documentsFinding answers in very long texts10Claude Sonnet 41 of 1 test
- ReasoningHard problems that need careful thinking13DeepSeek-R1 05283 of 4 tests
- FactsGetting facts right instead of making them up11Even1 each
Long documents rests on a single test.
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 76% less for the same work.
The biggest differences
The three tests each model wins by the widest margin. Scores are out of 100.
Where Claude Sonnet 4 pulls ahead
- Avoids making up answers it doesn't knowAA-Omniscience · Non-hallucination+54.3points ahead
- Fixes real GitHub issues in Python projectsSWE-bench Verified+28.1points ahead
- Fixes real GitHub issues in many programming languagesSWE-bench Multilingual+26.4points ahead
Where DeepSeek-R1 0528 pulls ahead
- Answers hard knowledge questions correctlyAA-Omniscience · Accuracy+7.8points ahead
- Very hard expert questions across many subjectsHumanity's Last Exam+5.1points ahead
- Graduate-level biology, physics and chemistry questionsGPQA Diamond+3.6points ahead
Every test, side by side
All 43 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingClaude Sonnet 4
- SWE-bench VerifiedClaude Sonnet 4 by 28.172.744.6+28.1
- SWE-bench MultilingualClaude Sonnet 4 by 26.456.930.5+26.4
- Terminal-Bench HardClaude Sonnet 4 by 15.231.115.9+15.2
- SciCodetie4040.3tie
ReasoningDeepSeek-R1 0528
- Humanity's Last ExamDeepSeek-R1 0528 by 5.110.715.8+5.1
- SimpleBenchClaude Sonnet 4 by 4.745.540.8+4.7
- GPQA DiamondDeepSeek-R1 0528 by 3.677.781.3+3.6
- CritPtDeepSeek-R1 0528 by 1.10.31.4+1.1
FactsEven
- AA-Omniscience · Non-hallucinationClaude Sonnet 4 by 54.370.916.6+54.3
- AA-Omniscience · AccuracyDeepSeek-R1 0528 by 7.822.730.5+7.8
Following instructionsClaude Sonnet 4
- IFBenchClaude Sonnet 4 by 15.154.739.6+15.1
- Multi-ChallengeClaude Sonnet 4 by 12.157.145+12.1
Other results28 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 76.415.992.3+76.4
- HMMT 2025DeepSeek-R1 0528 by 63.515.979.4+63.5
- AIME 2024DeepSeek-R1 0528 by 4843.491.4+48
- HarmBenchClaude Sonnet 4 by 43.49854.6+43.4
- livecodebench_hardDeepSeek-R1 0528 by 3132.663.6+31
- Terminal-BenchClaude Sonnet 4 by 29.835.55.7+29.8
- τ²-Bench Telecom (AA run)Claude Sonnet 4 by 28.164.636.5+28.1
- AA-OmniscienceClaude Sonnet 4 by 27.60.2-27.4+27.6
- ZebraLogicDeepSeek-R1 0528 by 21.473.795.1+21.4
- ARC-AGI-1Claude Sonnet 4 by 18.84021.2+18.8
- AIME 2025DeepSeek-R1 0528 by 1770.587.5+17
- TAU-bench (retail)Claude Sonnet 4 by 16.680.563.9+16.6
- livecodebench_mediumDeepSeek-R1 0528 by 14.97690.9+14.9
- Artificial Analysis Coding IndexClaude Sonnet 4 by 13.637.624+13.6
- BrowseComp-ZHDeepSeek-R1 0528 by 6.629.135.7+6.6
- TAU-bench (airline)Claude Sonnet 4 by 6.56053.5+6.5
- AA IntelligenceClaude Sonnet 4 by 5.818.913.1+5.8
- LiveCodeBenchDeepSeek-R1 0528 by 4.868.573.3+4.8
- MATH-500 (EM)DeepSeek-R1 0528 by 49498+4
- AA Agentic IndexDeepSeek-R1 0528 by 3.217.620.8+3.2
- XSTestDeepSeek-R1 0528 by 2.296.698.8+2.2
- simple_safety_testsClaude Sonnet 4 by 1.710098.3+1.7
- MMLU-ProDeepSeek-R1 0528 by 1.383.785+1.3
- Aider-Polyglottie70.771.6tie
- livecodebench_easytie98.899.2tie
- MMLU-Reduxtie93.693.4tie
- anthropic_red_teamtie99.299.1tie
- bbqtie96.696.5tie
Questions people ask
Which is better, Claude Sonnet 4 or DeepSeek-R1 0528?
Claude Sonnet 4 wins four of the six areas where both have results: coding, agents, long documents and following instructions. DeepSeek-R1 0528 wins reasoning, and costs 76% less. They are level on facts.
Which is better for coding?
Claude Sonnet 4. It wins 3 of the 4 coding tests both models report; DeepSeek-R1 0528 wins none, and 1 is a tie.
Which is cheaper?
Claude Sonnet 4 costs $3.00 per million input tokens and $15.00 per million output tokens; DeepSeek-R1 0528 costs $1.35 and $3.00. That makes DeepSeek-R1 0528 about 76% cheaper for the same work.
How do you compare the two?
We use the 43 benchmark tests both models have published scores on. The verdict counts the 15 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 28 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.