Claude Sonnet 5 vs DeepSeek-V4-Flash
Wins 6 of 7 areas
Coding · Agents · Reasoning · Facts · Math · Long documents
Wins 1 of 7 areas
Following instructions
Claude Sonnet 5 is the stronger all-rounder.DeepSeek-V4-Flash is cheaper and better at following instructions.
Scores updated · 37 tests both models report · How we compare
Where each one wins
Tests won in each of the seven 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 software80Claude Sonnet 58 of 8 tests
- ReasoningHard problems that need careful thinking30Claude Sonnet 53 of 5 tests · 2 ties
- MathCompetition and research-level math30Claude Sonnet 53 of 3 tests
- AgentsCarrying out multi-step tasks on its own31Claude Sonnet 53 of 4 tests
- FactsGetting facts right instead of making them up10Claude Sonnet 51 of 3 tests · 2 ties
- Long documentsFinding answers in very long texts10Claude Sonnet 51 of 1 test
- Following instructionsDoing exactly what it is asked01DeepSeek-V4-Flash1 of 1 test
Long documents and following instructions rest on a single test each.
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-V4-Flash costs 85% 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 Sonnet 5 pulls ahead
- Avoids making up answers it doesn't knowAA-Omniscience · Non-hallucination+52.3points ahead
- Common-sense trick questionsSimpleBench+14.3points ahead
- Fixes real GitHub issues, working as an agentLiveBench · Agentic Coding+12.6points ahead
Where DeepSeek-V4-Flash pulls ahead
- Customer-service tasks in a simulated bankτ-Bench V3 · Banking+2.1points ahead
- Follows detailed instructions exactlyLiveBench · Instruction Following+1.6points ahead
Every test, side by side
All 37 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingClaude Sonnet 5
- LiveBench · Agentic CodingClaude Sonnet 5 by 12.659.446.8+12.6
- LMArena · WebDevClaude Sonnet 5 by 111 rating points15411430+111 rating
- SWE-bench ProClaude Sonnet 5 by 10.663.252.6+10.6
- SWE-bench VerifiedClaude Sonnet 5 by 6.285.279+6.2
- LiveBench · CodingClaude Sonnet 5 by 5.780.775+5.7
- SWE-bench MultilingualClaude Sonnet 5 by 578.373.3+5
- SciCodeClaude Sonnet 5 by 454.350.3+4
- Terminal-Bench 2.1Claude Sonnet 5 by 1.880.578.7+1.8
AgentsClaude Sonnet 5
- BrowseCompClaude Sonnet 5 by 11.584.773.2+11.5
- τ-Bench V3 · BankingDeepSeek-V4-Flash by 2.137.339.4+2.1
- Terminal-Bench 4.0Claude Sonnet 5 by 214.112.1+2
- GDPValClaude Sonnet 5 by 1.448.346.9+1.4
ReasoningClaude Sonnet 5
- SimpleBenchClaude Sonnet 5 by 14.360.646.3+14.3
- Humanity's Last ExamClaude Sonnet 5 by 2.741.338.6+2.7
- LiveBench · ReasoningClaude Sonnet 5 by 2.188.786.6+2.1
- CritPttie16.916.6tie
- GPQA Diamondtie91.190.8tie
FactsClaude Sonnet 5
- AA-Omniscience · Non-hallucinationClaude Sonnet 5 by 52.360.68.3+52.3
- SimpleQA Verifiedtie33.734.1tie
- AA-Omniscience · Accuracytie4040.4tie
MathClaude Sonnet 5
- FrontierMath Tiers 1-3 (v2)Claude Sonnet 5 by 8.165.657.5+8.1
- LiveBench · MathematicsClaude Sonnet 5 by 6.192.986.8+6.1
- FrontierMath Tier 4Claude Sonnet 5 by 4.929.324.4+4.9
Following instructionsDeepSeek-V4-Flash
- LiveBench · Instruction FollowingDeepSeek-V4-Flash by 1.663.965.5+1.6
Other results12 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.
- GDPval-AA v2Claude Sonnet 5 by 429 rating points16181189+429 rating
- AA-OmniscienceClaude Sonnet 5 by 30.816.4-14.3+30.8
- Terminal-Bench 2.0Claude Sonnet 5 by 23.580.456.9+23.5
- ToolathlonDeepSeek-V4-Flash by 1654.370.3+16
- HLE (with tools)Claude Sonnet 5 by 12.357.445.1+12.3
- livebench_data_analysisDeepSeek-V4-Flash by 7.671.779.3+7.6
- Toolathlon VerifiedClaude Sonnet 5 by 4.474.770.3+4.4
- livebench_languageDeepSeek-V4-Flash by 4.27579.2+4.2
- AA IntelligenceClaude Sonnet 5 by 3.938.234.3+3.9
- AA Agentic IndexClaude Sonnet 5 by 2.644.341.7+2.6
- Artificial Analysis Coding IndexClaude Sonnet 5 by 2.471.569.1+2.4
- DeepSWE 1.1tie5454.4tie
Questions people ask
Which is better, Claude Sonnet 5 or DeepSeek-V4-Flash?
Claude Sonnet 5 wins six of the seven areas where both have results: coding, agents, reasoning, facts, math and long documents. DeepSeek-V4-Flash wins following instructions, and costs 85% less.
Which is better for coding?
Claude Sonnet 5. It wins 8 of the 8 coding tests both models report; DeepSeek-V4-Flash wins none.
Which is cheaper?
Claude Sonnet 5 costs $2.00 per million input tokens and $10.00 per million output tokens; DeepSeek-V4-Flash costs $0.44 and $1.32. That makes DeepSeek-V4-Flash about 85% cheaper for the same work.
How do you compare the two?
We use the 37 benchmark tests both models have published scores on. The verdict counts the 25 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 12 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.