Atlas

Compare Models

Claude Opus 5 vs DeepSeek V4.1 Flash

Claude Opus 5 leads by 4.7 capability points (≈95% interval +2.5 to +6.8): a clear lead over DeepSeek V4.1 Flash · P(higher) ≈ 100%

Wins 171 of 220 shared benchmarks (28 lost, 21 tied)

Capability vs. relative cost
Claude Opus 5DeepSeek V4.1 FlashClosed weightsOpen weights

Each line runs through one model’s priced reasoning settings, cheapest to most expensive; grey points are other models’ priced settings in the same range. Results whose reasoning effort the source didn’t report are left out wherever a model has named effort levels. Capability and relative cost are the published estimates (relative cost is cost per task as a multiple of GPT-5 (high)); hover a setting, or use the arrow keys, to set it beside the other model’s same setting.

Claude Opus 5vsDeepSeek V4.1 Flashcapability vs. relative costData Nov 15, 2026Village LLM Atlasatlas.village.dev

Benchmark wins171 – 28 · 220 shared, 21 tied

Comparing matching setups only (20 benchmarks excluded) · suites counted once
← Claude Opus 5DeepSeek V4.1 Flash →

Claude Opus 5vsDeepSeek V4.1 Flashwins by categoryData Nov 15, 2026Data: Vals AI, Arena, benchmarks.bio and 52 othersVillage LLM Atlasatlas.village.dev

Margins

Each benchmark scaled 0–100% (its declared range, else the field’s range); Δ is DeepSeek V4.1 Flash minus Claude Opus 5.

BrokenArXiv 06/2026Math90.739.4−51.4
LLM Creative Story WritingChat & Writing3.7-0.5−47.7
FrontierMath Tier 4 (v2) - PrivateMath73.226.8−46.3
ProofBench v1.1Math99.054.0−45.0
IOI (Vals v2)Code84.340.3−44.1
RiemannBenchMath68.024.0−44.0
Medical Long Context Reasoning (MLCR-AA)Health & Medicine59.422.8−36.7
KernelBench Hard Normalized (RTX PRO 6000)Code33.369.7+36.4
Creative Writing v3 EloChat & Writing21351540−30.1
ARC-AGI-3 (Semi-Private)General QA30.20.5−29.7
A-Fantasia (Valid Attempts with Retries)Games29.759.3−29.7
ALE-BenchCode26221829−28.1
EnterpriseBench: CoreCraftProfessional Work68.741.5−27.2
Terminal-Bench Science 0.1 (Vals)Science27.10.0−27.1
Furniture Assembly BenchmarkMultimodal60.834.2−26.7
APEX-Agents 1.1Professional Work77.050.5−26.5
Humanity's Last Exam DiamondGeneral QA44.819.1−25.7
OncologyBench Strict (25 tasks, October 9 rubric)Health & Medicine40.014.7−25.3
ORCA-bench Hard RCA (Fireworks 324-task subset)Agents44.320.3−24.1
Terminal-Bench 4.0Code53.931.2−22.7
MedCodeHealth & Medicine63.641.2−22.4
Conceptual Reasoning IndexIndexes74.758.4−22.3
Terminal-Bench 4.0 (Artificial Analysis)Code49.026.8−22.2
BuseyBench SVGMultimodal6.34.1−22.1
AA-Omniscience IndexIndexes37.1-5.3−21.2
Wiz Cyber Model Arena v3.9Cybersecurity71.250.2−21.0
BizAgentBench (strict success)Agents64.043.3−20.7
HealthBench ProfessionalHealth & Medicine73.452.9−20.5
LLM Debate BenchmarkChat & Writing17241582−20.2
GBENCH One-shot Coding GScoreGames68.148.0−20.2
Harvey's Legal Agent Benchmark - Data Privacy CybersecurityLegal0.020.0+20.0
Harvey's Legal Agent Benchmark - Structured Finance SecuritizationLegal20.00.0−20.0
Harvey's Legal Agent Benchmark - Trusts Estates Private ClientLegal0.020.0+20.0
Humanity's Last ExamGeneral QA56.436.8−19.6
Terminal-Bench-Science 0.1Science28.69.0−19.5
VariantBench V1 Clean RerunsScience50.631.1−19.5
LMArena Agent Arena Chat Net Improvement (Steering Burden)Agents9.61.6−19.3
AA-AnalystAgent (pass^5)Professional Work53.835.0−18.7
LMArena Agent Arena Net ImprovementAgents9.53.9−18.7
Vals Science IndexIndexes47.929.6−18.3
FrontierMath Tiers 1-3 (v2) - PrivateMath85.667.4−18.2
SUPERChemScience74.756.7−18.0
ARC-AGI-2General QA90.472.9−17.5
LMCAGeneral QA64.547.0−17.5
ProgramBench - almost solved rollout percentageCode37.020.3−16.7
GameBench 2Games84.868.4−16.4
Excel Modeling BenchmarkFinance & Accounting73.657.2−16.3
TxBench Preclinical Pharmacology V1 Clean RerunsScience63.046.7−16.3
MedXpertQA MM (200-question set)Health & Medicine79.863.5−16.3
MysteryMechanismScience37.421.2−16.2
BullshitBench v2 LeaderboardSafety73.057.0−16.0
MetagenomicsBench V0Science54.038.0−16.0
MetagenomicsBench V1Science54.038.0−16.0
Mystery Game PuzzlesGames59.043.0−16.0
TaxCalcBench TY25Finance & Accounting20.04.0−16.0
ACCoRDGeneral QA83.868.1−15.8
HLE Extended (Mercor held-out)General QA61.045.3−15.7
MedXpertQA MM Extended (Mercor held-out)Health & Medicine64.148.4−15.7
Concept Synth INDUCTION Challenge64 Round 1General QA32.817.2−15.6
CritPt (avg@5)Science29.114.3−14.9
Tax Agent BenchFinance & Accounting76.762.5−14.2
Legal Research BenchLegal55.341.3−13.9
SimpleBenchGeneral QA80.666.7−13.9
τ³-Banking (Fireworks, banking_knowledge)Agents49.135.7−13.4
Terminal-Bench 3 - release-reported accuracyCode43.330.0−13.3
BioSecBench-Refusal V1 Clean RerunsSafety24.511.3−13.2
WeirdML v3Science19.05.9−13.1
Concept Synth INDUCTION Challenge100 Round 1General QA24.011.0−13.0
AA-Briefcase EloAgents16451424−12.9
LMArena Agent Arena Work Net Improvement (Steering Burden)Agents7.82.5−12.8
Public Benefits Bench v1.1Professional Work76.964.3−12.7
ScBench-Long V1 Clean RerunsScience27.314.8−12.5
ComplexConstraintsChat & Writing37.349.7+12.4
TxBench Antibody Discovery V0Science51.439.0−12.4
AutomationBench-AA 1.0.6Agents56.668.9+12.3
Longform Creative WritingChat & Writing86.374.0−12.3
BioSecBench-Structure V1 (58 evaluations)Science15.527.6+12.1
LMArena Agent Arena Code Net Improvement (Steering Burden)Agents8.02.7−12.0
Senior SWE-BenchCode34.722.7−12.0
dfbench Detection RecallCybersecurity45.457.3+11.9
Code MigrationCode57.545.6−11.9
APEX-SWECode63.751.9−11.8
VIALSMultimodal24.412.6−11.8
Bedside BenchHealth & Medicine94.682.9−11.7
Arena Image-to-WebDevCode16641610−11.7
BioMysteryBench (Vals Terminus 2)Science79.367.8−11.5
Artificial Analysis Intelligence Index v4.3.2Indexes50.839.5−11.3
NL2Repo - release-reported repository generation scoreCode75.364.0−11.3
MineBench — Bradley-Terry RatingGames20081822−11.3
GDPval-AA v2Professional Work18611632−11.3
Artificial Analysis Intelligence Index v4.3Indexes50.739.5−11.2
Creative Writing v3 Rubric ScoreChat & Writing16.314.1−11.1
GeneBench-Pro Public Release SubsetScience22.011.0−11.0
CharXiv Extended (Mercor held-out)Multimodal79.868.9−10.9
DAXBench v2Code86.275.5−10.6
ReactBench (39 tasks): Writing ReactCode57.447.0−10.4
DistressBenchSafety70.279.8+9.6
SpeciEval (Pooled English Runs)Safety87.997.4+9.4
SpeciEval (Refusal-Aware Retries, Historical)Safety87.997.4+9.4
BioSecBench-RespiratoryVirus-Surveillance V1 (110 evaluations)Science14.65.2−9.4
MMMU-Pro Extended (Mercor held-out)Multimodal34.325.0−9.3
LMArena Code Arena WebDevCode16911621−9.2
AutoformBenchMath61.952.8−9.1
ScBench-Long V0 VerifiedScience36.427.3−9.1
TxBench Antibody Discovery V1 (100 evaluations)Science48.739.7−9.0
YC-BenchGames19115921657236−8.9
SpatialBench V0 VerifiedScience66.775.4+8.7
BigFinanceBench private holdout (final-answer accuracy)Finance & Accounting49.240.6−8.6
TxBench Oligo Discovery V0Science42.233.9−8.3
LLM Chess EloGames12851087−8.1
DTBenchGeneral QA97.989.9−8.0
TxBench Oligo Discovery V1 (113 evaluations)Science42.534.8−7.7
FrontierCode 1.1 Extended - Pass RateCode69.662.0−7.6
LMArena Agent Arena Chat Praise vs ComplaintAgents18.23.1−7.6
LMArena Agent Arena Chat Bash ReliabilityAgents13.7-1.0−7.4
BlenderWorldCode73.065.7−7.3
LMArena Agent Arena SteerabilityAgents10.7-3.3−7.0
EpiBenchScience35.242.1+6.9
ExploitGym - release-reported percentage, cohort unspecifiedCybersecurity22.115.3−6.8
Epoch Capabilities Index (ECI)Indexes164156−6.5
APEX-AccountingFinance & Accounting54.548.1−6.4
τ³-Banking (Mercor run, bm25_grep retrieval)Agents44.037.8−6.2
SciCode (main problems)Science17.311.2−6.1
Vision Arena OverallMultimodal12851263−5.8
GDPval-AA v2.1Professional Work17291600−5.7
MedScribeHealth & Medicine91.085.5−5.5
LMArena Agent Arena Work Bash ReliabilityAgents10.90.1−5.4
DeepsecBench-AA (Artificial Analysis)Cybersecurity25.931.2+5.3
HealthBenchHealth & Medicine67.161.9−5.2
SpatialBench V1 Clean RerunsScience68.463.2−5.2
Aikido Known-CVE Detection (32 CVEs) single-run averageCybersecurity70.865.6−5.2
BrowseComp (Mercor 130-question subset)Search84.679.4−5.2
LMArena Agent Arena Code Bash ReliabilityAgents10.0-0.2−5.1
Chartography (with tools)Multimodal84.078.9−5.1
Bug Hunt BenchCode25.720.6−5.1
Chess Puzzles (task version 2.0)Games43.048.0+5.0
SWEeper-BenchCode56.551.5−5.0
Political Science LLM BenchmarkProfessional Work71.466.5−4.9
Artificial Analysis Progress Index v1Indexes334320−4.7
Extended NYT Connections 940 (Quadratic v1)Games94.389.6−4.7
SciCode 1.0.1 (Artificial Analysis)Code56.451.9−4.5
BabyVision with toolsMultimodal94.189.6−4.5
CharXiv (descriptive and reasoning combined)Multimodal94.690.1−4.5
AutomationBench 1.0.6 (provider-reported)Agents50.354.8+4.5
BioSecBench-Function V0Science43.038.7−4.3
GDP.pdfProfessional Work24.019.8−4.2
LiveBench 2.0Agents82.178.0−4.0
CadQueryEval (Corrected Geometry)Code80.076.0−4.0
LMArena Agent Arena Chat Confirmed SuccessAgents12.84.9−4.0
LMArena Agent Arena Work Praise vs ComplaintAgents11.63.7−3.9
LMArena Agent Arena Code Steering BurdenAgents10.12.3−3.9
MMLU ProGeneral QA92.088.2−3.8
LegalBenchLegal87.083.3−3.7
RCM-Bench-1Health & Medicine92.789.0−3.7
Vibe Code Bench v1.1Code88.484.7−3.7
ScBench V1 Clean RerunsScience58.354.7−3.6
AdvancedIFChat & Writing80.584.0+3.5
SWE Atlas Codebase QnA Extended (Mercor held-out)Code40.036.7−3.3
Agents' Last Exam - release-reported pass rate, subset unspecifiedAgents28.631.8+3.2
AA-LCR Extended (Mercor held-out)Search73.970.8−3.1
APEX-1 (Mercor, mean score)Professional Work68.665.5−3.1
SpatialBench-Long V0 VerifiedScience31.834.9+3.0
ARC-AGI-1 Semi-PrivateGeneral QA97.594.5−3.0
ZeroBench (Main questions, pass@5, with tools)Multimodal52.049.0−3.0
DeepResearch Bench II (Mercor 130-task set)Search56.153.2−2.9
LMArena Agent Arena Work Steering BurdenAgents7.31.5−2.9
LMArena Agent Arena Code Praise vs ComplaintAgents8.83.0−2.9
LMArena Agent Arena Bash RecoveryAgents13.67.9−2.9
SPU (Science Prediction Universe)Science0.20.2+2.7
LMArena Text Arena EloChat & Writing14901475−2.6
LMArena Agent Arena Chat Steering BurdenAgents4.6-0.5−2.5
LMArena Agent Arena Code Confirmed SuccessAgents11.66.6−2.5
LMArena Agent Arena Work Confirmed SuccessAgents10.25.7−2.3
SpatialBench-Long V1 Clean RerunsScience28.026.2−1.8
AIME 2025Math100.098.3−1.7
Harvey LAB: In-House Contracting (held-out)Legal2.00.3−1.7
Harvey LAB held-out set (Fireworks run)Legal7.45.7−1.7
BioSecBench-Function V1 Clean RerunsScience42.540.8−1.6
SAGEMath49.447.9−1.5
KernelBench CUDA Normalized (RTX PRO 6000)Code45.547.0+1.5
Terminal-Bench 2.1Code89.190.6+1.5
Epiq LAER CorporateBenchProfessional Work74.075.3+1.3
ReactBench (39 tasks): Fixing ReactCode20.018.8−1.2
GDPval (Mercor 213-task rubric score)Professional Work85.686.7+1.1
DeepSWE v1.1Code73.274.2+1.0
Eval Connections One-Shot with TrapsGames74.073.0−1.0
Design Arena Overall Code CategoriesCode13211316−0.8
Arena Alignment IndexSafety75.674.8−0.8
RuneBenchGames5.75.7−0.8
SWE-bench Verified Extended (Mercor held-out)Code82.082.7+0.7
MacroscopeBenchCode69.970.5+0.6
OTIS Mock AIME 2024-2025Math98.998.3−0.6
Terminal-Bench 2.1 Extended (Mercor held-out)Code38.038.4+0.4
EpiBench V1 Clean RerunsScience25.225.5+0.3
Harvey HLAB — Criteria Pass RateLegal87.787.5−0.2
LMArena Agent Arena Work Tool HallucinationAgents0.70.5−0.1
LMArena Agent Arena Chat Tool HallucinationAgents0.50.3−0.1
LMArena Agent Arena Code Tool HallucinationAgents0.70.5−0.1
BioSecBench-Surveillance V1 Clean RerunsScience51.551.50.0
Harvey's Legal Agent Benchmark - Antitrust CompetitionLegal0.00.00.0
Harvey's Legal Agent Benchmark - Arbitration International Dispute ResolutionLegal0.00.00.0
Harvey's Legal Agent Benchmark - Banking FinanceLegal0.00.00.0
Harvey's Legal Agent Benchmark - Bankruptcy RestructuringLegal0.00.00.0
Harvey's Legal Agent Benchmark - Capital MarketsLegal0.00.00.0
Harvey's Legal Agent Benchmark - Corporate GovernanceLegal0.00.00.0
Harvey's Legal Agent Benchmark - Corporate M&ALegal20.020.00.0
Harvey's Legal Agent Benchmark - Emerging Companies Venture CapitalLegal0.00.00.0
Harvey's Legal Agent Benchmark - Employment LaborLegal0.00.00.0
Harvey's Legal Agent Benchmark - Energy Natural ResourcesLegal20.020.00.0
Harvey's Legal Agent Benchmark - Environmental ESGLegal0.00.00.0
Harvey's Legal Agent Benchmark - Funds Asset ManagementLegal0.00.00.0
Harvey's Legal Agent Benchmark - Healthcare Life SciencesLegal20.020.00.0
Harvey's Legal Agent Benchmark - ImmigrationLegal0.00.00.0
Harvey's Legal Agent Benchmark - InsuranceLegal0.00.00.0
Harvey's Legal Agent Benchmark - Intellectual PropertyLegal0.00.00.0
Harvey's Legal Agent Benchmark - International Trade SanctionsLegal0.00.00.0
Harvey's Legal Agent Benchmark - Litigation Dispute ResolutionLegal0.00.00.0
Harvey's Legal Agent Benchmark - Real EstateLegal0.00.00.0
Harvey's Legal Agent Benchmark - TaxLegal0.00.00.0
Harvey's Legal Agent Benchmark - White Collar Defense InvestigationsLegal0.00.00.0

Claude Opus 5vsDeepSeek V4.1 Flashnormalised score marginsData Nov 15, 2026Data: Vals AI, Arena, benchmarks.bio and 52 othersVillage LLM Atlasatlas.village.dev

Win counts use each benchmark’s declared score direction. Represented suite slices and the derived Capability score are excluded from headline totals. Missing or incompatible setups are excluded, never counted as losses.
Loading Atlas data…