Claude Sonnet 4.5 vs DeepSeek-V4-Flash
Wins 0 of 7 areas
—
Wins 7 of 7 areas
Coding · Agents · Reasoning · Facts · Math · Long documents · Following instructions
DeepSeek-V4-Flash is the stronger all-rounder.
Scores updated · 39 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 software07DeepSeek-V4-Flash7 of 8 tests · 1 tie
- AgentsCarrying out multi-step tasks on its own05DeepSeek-V4-Flash5 of 5 tests
- ReasoningHard problems that need careful thinking14DeepSeek-V4-Flash4 of 5 tests
- MathCompetition and research-level math03DeepSeek-V4-Flash3 of 3 tests
- FactsGetting facts right instead of making them up12DeepSeek-V4-Flash2 of 3 tests
- Long documentsFinding answers in very long texts01DeepSeek-V4-Flash1 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 90% 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 DeepSeek-V4-Flash pulls ahead
- Finds hard-to-locate facts by browsing the webBrowseComp+49.1points ahead
- Abstract visual puzzles that people can solveARC-AGI-2+47.8points ahead
- Unpublished advanced math problemsFrontierMath Tiers 1-3 (v2)+33.6points ahead
Where Claude Sonnet 4.5 pulls ahead
- Avoids making up answers it doesn't knowAA-Omniscience · Non-hallucination+42.6points ahead
- Common-sense trick questionsSimpleBench+8points 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.
CodingDeepSeek-V4-Flash
- LiveCodeBench v6DeepSeek-V4-Flash by 26.66490.6+26.6
- Terminal-Bench 2.1DeepSeek-V4-Flash by 22.955.878.7+22.9
- SWE-bench ProDeepSeek-V4-Flash by 943.652.6+9
- SWE-bench MultilingualDeepSeek-V4-Flash by 6.36773.3+6.3
- SciCodeDeepSeek-V4-Flash by 4.645.750.3+4.6
- LMArena · WebDevDeepSeek-V4-Flash by 36 rating points13941430+36 rating
- SWE-bench VerifiedDeepSeek-V4-Flash by 1.877.279+1.8
- Terminal-Bench Hardtie35.635.6tie
AgentsDeepSeek-V4-Flash
- BrowseCompDeepSeek-V4-Flash by 49.124.173.2+49.1
- GDPValDeepSeek-V4-Flash by 26.420.546.9+26.4
- MCP AtlasDeepSeek-V4-Flash by 25.243.869+25.2
- τ-Bench V3 · BankingDeepSeek-V4-Flash by 14.924.539.4+14.9
- Terminal-Bench 4.0DeepSeek-V4-Flash by 12.1012.1+12.1
ReasoningDeepSeek-V4-Flash
- ARC-AGI-2DeepSeek-V4-Flash by 47.813.661.4+47.8
- Humanity's Last ExamDeepSeek-V4-Flash by 20.817.838.6+20.8
- CritPtDeepSeek-V4-Flash by 15.51.116.6+15.5
- SimpleBenchClaude Sonnet 4.5 by 854.346.3+8
- GPQA DiamondDeepSeek-V4-Flash by 7.483.490.8+7.4
FactsDeepSeek-V4-Flash
- AA-Omniscience · Non-hallucinationClaude Sonnet 4.5 by 42.650.98.3+42.6
- AA-Omniscience · AccuracyDeepSeek-V4-Flash by 7.532.940.4+7.5
- SimpleQA VerifiedDeepSeek-V4-Flash by 3.430.734.1+3.4
MathDeepSeek-V4-Flash
- FrontierMath Tiers 1-3 (v2)DeepSeek-V4-Flash by 33.623.957.5+33.6
- IMOAnswerBenchDeepSeek-V4-Flash by 22.565.988.4+22.5
- FrontierMath Tier 4DeepSeek-V4-Flash by 222.424.4+22
Long documentsDeepSeek-V4-Flash
- AA-LCRDeepSeek-V4-Flash by 7.472.379.7+7.4
Following instructionsDeepSeek-V4-Flash
- IFBenchDeepSeek-V4-Flash by 21.957.379.2+21.9
Other results13 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.
- BrowseComp (context management)DeepSeek-V4-Flash by 47.126.173.2+47.1
- ToolathlonDeepSeek-V4-Flash by 29.34170.3+29.3
- ARC-AGI-1DeepSeek-V4-Flash by 25.363.789+25.3
- AA Agentic IndexDeepSeek-V4-Flash by 24.217.541.7+24.2
- LiveCodeBenchDeepSeek-V4-Flash by 20.67191.6+20.6
- Artificial Analysis Coding IndexDeepSeek-V4-Flash by 1752.169.1+17
- τ²-Bench Telecom (AA run)DeepSeek-V4-Flash by 16.978.195+16.9
- AA-OmniscienceClaude Sonnet 4.5 by 14.2-0.1-14.3+14.2
- AA IntelligenceDeepSeek-V4-Flash by 13.620.734.3+13.6
- HLE (with tools)DeepSeek-V4-Flash by 13.13245.1+13.1
- Terminal-Bench 2.0DeepSeek-V4-Flash by 6.95056.9+6.9
- GPQA (unspecified)DeepSeek-V4-Flash by 5.183.488.5+5.1
- MMLU-ProClaude Sonnet 4.5 by 1.387.586.2+1.3
Questions people ask
Which is better, Claude Sonnet 4.5 or DeepSeek-V4-Flash?
DeepSeek-V4-Flash wins all seven areas where both have results: coding, agents, reasoning, facts, math, long documents and following instructions. Claude Sonnet 4.5 wins none.
Which is better for coding?
DeepSeek-V4-Flash. It wins 7 of the 8 coding tests both models report; Claude Sonnet 4.5 wins none, and 1 is a tie.
Which is cheaper?
Claude Sonnet 4.5 costs $3.00 per million input tokens and $15.00 per million output tokens; DeepSeek-V4-Flash costs $0.44 and $1.32. That makes DeepSeek-V4-Flash about 90% 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 26 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 13 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.