DeepSeek-V4-Flash vs Claude Opus 5
Wins 1 of 7 areas
Following instructions
Wins 5 of 7 areas
Coding · Agents · Reasoning · Facts · Math
Claude Opus 5 is the stronger all-rounder.DeepSeek-V4-Flash is cheaper and better at following instructions.
Scores updated · 41 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 software08Claude Opus 58 of 8 tests
- ReasoningHard problems that need careful thinking06Claude Opus 56 of 6 tests
- AgentsCarrying out multi-step tasks on its own05Claude Opus 55 of 5 tests
- FactsGetting facts right instead of making them up03Claude Opus 53 of 3 tests
- MathCompetition and research-level math03Claude Opus 53 of 3 tests
- Following instructionsDoing exactly what it is asked10DeepSeek-V4-Flash1 of 1 test
- Long documentsFinding answers in very long texts00Even0 each · 1 tie
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 94% 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 5 pulls ahead
- Extremely hard research-level math problemsFrontierMath Tier 4+48.8points ahead
- Complex command-line tasks across many fieldsTerminal-Bench 4.0+36.9points ahead
- Common-sense trick questionsSimpleBench+34.3points ahead
Where DeepSeek-V4-Flash pulls ahead
- Follows detailed instructions exactlyLiveBench · Instruction Following+1.7points ahead
Every test, side by side
All 41 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingClaude Opus 5
- SWE-bench ProClaude Opus 5 by 26.652.679.2+26.6
- LMArena · WebDevClaude Opus 5 by 265 rating points14301695+265 rating
- LiveBench · Agentic CodingClaude Opus 5 by 18.446.865.2+18.4
- SWE-bench VerifiedClaude Opus 5 by 177996+17
- SWE-bench MultilingualClaude Opus 5 by 16.273.389.5+16.2
- Terminal-Bench 2.1Claude Opus 5 by 10.478.789.1+10.4
- LiveBench · CodingClaude Opus 5 by 6.57581.5+6.5
- SciCodeClaude Opus 5 by 6.150.356.4+6.1
AgentsClaude Opus 5
- Terminal-Bench 4.0Claude Opus 5 by 36.912.149+36.9
- BrowseCompClaude Opus 5 by 17.673.290.8+17.6
- MCP AtlasClaude Opus 5 by 16.86985.8+16.8
- GDPValClaude Opus 5 by 14.346.961.2+14.3
- τ-Bench V3 · BankingClaude Opus 5 by 2.739.442.1+2.7
ReasoningClaude Opus 5
- SimpleBenchClaude Opus 5 by 34.346.380.6+34.3
- ARC-AGI-2Claude Opus 5 by 2961.490.4+29
- Humanity's Last ExamClaude Opus 5 by 16.338.654.9+16.3
- CritPtClaude Opus 5 by 12.516.629.1+12.5
- LiveBench · ReasoningClaude Opus 5 by 4.686.691.2+4.6
- GPQA DiamondClaude Opus 5 by 2.490.893.2+2.4
FactsClaude Opus 5
- AA-Omniscience · Non-hallucinationClaude Opus 5 by 30.98.339.2+30.9
- SimpleQA VerifiedClaude Opus 5 by 25.834.159.9+25.8
- AA-Omniscience · AccuracyClaude Opus 5 by 20.540.460.9+20.5
MathClaude Opus 5
- FrontierMath Tier 4Claude Opus 5 by 48.824.473.2+48.8
- FrontierMath Tiers 1-3 (v2)Claude Opus 5 by 28.157.585.6+28.1
- LiveBench · MathematicsClaude Opus 5 by 8.986.895.7+8.9
Following instructionsDeepSeek-V4-Flash
- LiveBench · Instruction FollowingDeepSeek-V4-Flash by 1.765.563.8+1.7
Other results14 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 Opus 5 by 635 rating points11891824+635 rating
- AA-OmniscienceClaude Opus 5 by 51.4-14.337.1+51.4
- GDPval-AA (Elo)Claude Opus 5 by 313 rating points13951708+313 rating
- JobBenchClaude Opus 5 by 24.441.365.7+24.4
- HLE (with tools)Claude Opus 5 by 18.545.163.6+18.5
- AA IntelligenceClaude Opus 5 by 16.534.350.8+16.5
- AA Agentic IndexClaude Opus 5 by 14.541.756.2+14.5
- DeepSWE 1.1Claude Opus 5 by 14.454.468.8+14.4
- Toolathlon VerifiedClaude Opus 5 by 10.370.380.6+10.3
- livebench_languageClaude Opus 5 by 9.579.288.7+9.5
- Artificial Analysis Coding IndexClaude Opus 5 by 8.969.178+8.9
- ARC-AGI-1Claude Opus 5 by 8.58997.5+8.5
- Agents' Last ExamClaude Opus 5 by 6.425.231.6+6.4
- livebench_data_analysisDeepSeek-V4-Flash by 4.779.374.5+4.7
Questions people ask
Which is better, DeepSeek-V4-Flash or Claude Opus 5?
Claude Opus 5 wins five of the seven areas where both have results: coding, agents, reasoning, facts and math. DeepSeek-V4-Flash wins following instructions, and costs 94% less. They are level on long documents.
Which is better for coding?
Claude Opus 5. It wins 8 of the 8 coding tests both models report; DeepSeek-V4-Flash wins none.
Which is cheaper?
DeepSeek-V4-Flash costs $0.44 per million input tokens and $1.32 per million output tokens; Claude Opus 5 costs $5.00 and $25.00. That makes DeepSeek-V4-Flash about 94% cheaper for the same work.
How do you compare the two?
We use the 41 benchmark tests both models have published scores on. The verdict counts the 27 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 14 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.