Atlas

Compare Models

DeepSeek V4.1 Flash vs GPT-6 Astra

GPT-6 Astra leads by 7.8 capability points (≈95% interval +5.5 to +10.2): a clear lead over DeepSeek V4.1 Flash · P(higher) ≈ 100%

Wins 181 of 252 shared benchmarks (48 lost, 23 tied)

Capability vs. relative cost
DeepSeek V4.1 FlashGPT-6 AstraClosed 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.

DeepSeek V4.1 FlashvsGPT-6 Astracapability vs. relative costData Nov 15, 2026Village LLM Atlasatlas.village.dev

Benchmark wins48 – 181 · 252 shared, 23 tied

Comparing matching setups only (87 benchmarks excluded) · suites counted once
← DeepSeek V4.1 FlashGPT-6 Astra →

DeepSeek V4.1 FlashvsGPT-6 Astrawins by categoryData Nov 15, 2026Data: Vals AI, Artificial Analysis, benchmarks.bio and 51 othersVillage LLM Atlasatlas.village.dev

Margins

Each benchmark scaled 0–100% (its declared range, else the field’s range); Δ is GPT-6 Astra minus DeepSeek V4.1 Flash.

RSI Arena Stage 1 human ratingAgents8871095+100.0
ARC-AGI-3 (Semi-Private)General QA3.299.9+96.8
Concept Synth INDUCTION Challenge100 Round 1General QA11.094.0+83.0
Concept Synth INDUCTION Challenge64 Round 1General QA17.298.4+81.3
FrontierMath Tier 4 (v2) - PrivateMath26.897.6+70.8
Design Creativity BenchCode43.665.5+69.5
Terminal-Bench Science 0.1 (Vals)Science0.062.9+62.9
BrokenArXiv 05/2026Math28.591.0+62.5
IOI (Vals v2)Code40.3100.0+59.7
BrokenArXiv 06/2026Math39.499.1+59.7
ChartographyMultimodal11.671.0+59.4
SRE Bench — Fully SolvedCode0.856.9+56.1
AA-PersonalityBench Conviction indexChat & Writing96.244.8−51.9
Terminal-Bench-Science 0.1Science9.059.5+50.5
AA-PersonalityBench Informativeness indexChat & Writing79.129.6−50.0
RiemannBenchMath24.072.0+48.0
AA-PersonalityBench Imagination indexChat & Writing31.577.2+46.2
ALE-BenchCode18293128+46.0
Furniture Assembly BenchmarkMultimodal34.280.0+45.8
ArXivMath 04/2026Math45.090.8+45.8
ProofBench v1.1Math54.099.0+45.0
Humanity's Last Exam DiamondGeneral QA19.163.8+44.7
LLM Creative Story WritingChat & Writing-0.53.4+44.4
Mystery Game PuzzlesGames43.084.0+41.0
TaxCalcBench TY25Finance & Accounting4.044.0+40.0
VIALSMultimodal12.649.9+37.3
ArXivMath 06/2026Math57.694.4+36.8
ArXivMath 05/2026Math58.395.0+36.7
WeirdML v3Science5.842.2+36.4
FrontierCUA pass@1Agents6.742.2+35.6
Vals Science IndexIndexes29.665.1+35.4
MMMU-Pro Extended (Mercor held-out)Multimodal25.060.0+35.0
AA-PersonalityBench Transparency indexChat & Writing60.094.3+34.6
AA-PersonalityBench Warmth indexChat & Writing41.075.3+34.6
RareDiseaseBench V1 (35 evaluations)Science0.033.3+33.3
Terminal-Bench 4.0 (Artificial Analysis)Code26.859.6+32.8
SimpleQA Verified (Epoch anti-abstention)General QA42.975.6+32.7
CyberBench v1.1 OverallCybersecurity73.741.1−32.6
BuseyBench SVGMultimodal4.17.3+32.3
Creative Writing v3 EloChat & Writing15402177+32.2
MysteryMechanismScience21.253.2+32.0
OpenWorldEngineer (Mean@8)Code12.444.2+31.8
GBENCH One-shot Coding GScoreGames48.079.2+31.3
AA-PersonalityBench Organization indexChat & Writing10069.5−30.8
Code2Video-AA EloCode10531427+30.3
ORCA-bench Hard RCA (Fireworks 324-task subset)Agents20.350.2+30.0
AutoformBenchMath52.882.4+29.6
BioSecBench-Refusal V1 Clean RerunsSafety11.340.8+29.5
LMArena Agent Arena Code Net Improvement (Steering Burden)Agents2.714.8+27.4
Terminal-Bench 4.0Code31.257.9+26.7
FrontierMath Tiers 1-3 (v2) - PrivateMath67.493.7+26.3
Decision-Bench v1.1Professional Work25.351.5+26.2
Chess Puzzles (task version 2.0)Games48.074.0+26.0
GameBench 2Games68.494.3+25.9
LMArena Agent Arena Work Net Improvement (Steering Burden)Agents2.513.2+25.8
GeneBench-Pro Public Release SubsetScience11.036.0+25.0
LMArena Agent Arena Chat Net Improvement (Steering Burden)Agents1.611.7+24.6
APEX-Agents 1.1Professional Work50.575.1+24.6
AA-Omniscience IndexIndexes-5.343.7+24.5
CadQueryEval (Corrected Geometry)Code76.0100.0+24.0
SUPERChemScience56.780.3+23.6
MedXpertQA MM (200-question set)Health & Medicine63.586.8+23.3
CyberGym-E2E-AA (Artificial Analysis)Cybersecurity22.90.0−22.9
SEC-Bench Pro - release-reported cohortCybersecurity62.885.4+22.6
Code MigrationCode45.667.7+22.1
ARC-AGI-2General QA72.995.0+22.1
Code Migration v1.1Code58.980.6+21.6
LLM Chess EloGames10871614+21.6
TxBench Oligo Discovery V0Science33.955.5+21.5
LMArena Agent Arena Net ImprovementAgents3.910.4+21.5
DAXBench v2Code75.596.8+21.3
Conceptual Reasoning IndexIndexes58.473.9+21.3
AA-PersonalityBench Slop indexChat & Writing37.216.2−21.2
Eval Connections One-Shot with TrapsGames73.094.0+21.0
CloudOpsBench v1.0Agents21.842.8+21.0
Harvey's Legal Agent Benchmark - Data Privacy CybersecurityLegal20.00.0−20.0
Harvey's Legal Agent Benchmark - Energy Natural ResourcesLegal20.00.0−20.0
Harvey's Legal Agent Benchmark - Healthcare Life SciencesLegal20.00.0−20.0
Wiz Cyber Model Arena v3.9Cybersecurity50.270.2+20.0
GDP.pdf (Artificial Analysis, All-pass)Professional Work12.832.2+19.4
OpenDS-WorldSearch20.539.0+18.5
TxBench-Antibody-Development V1 (60 evaluations)Science14.432.8+18.3
Humanity's Last ExamGeneral QA36.854.8+18.0
TxBench Antibody Discovery V0Science39.056.9+17.9
Real-SWE 1.1Code38.456.0+17.6
CritPt (avg@5)Science14.331.7+17.4
LMCAGeneral QA47.064.4+17.4
TxBench Oligo Discovery V1 (113 evaluations)Science34.851.9+17.1
TissueMicroscopyBench V1 (49 evaluations)Science2.719.7+17.0
SimpleBenchGeneral QA66.783.6+16.9
HealthBench ProfessionalHealth & Medicine52.969.7+16.7
TxBench Antibody Discovery V1 (100 evaluations)Science39.756.0+16.3
AA-AnalystAgent (pass^5)Professional Work35.051.3+16.2
CharXiv Extended (Mercor held-out)Multimodal68.985.1+16.2
VariantBench V1 Clean RerunsScience31.146.6+15.5
ScBench-Long V0 VerifiedScience27.342.4+15.2
Quant-Arena (iterative total score)Finance & Accounting36.551.3+14.8
BrowseComp (Mercor 130-question subset)Search79.494.2+14.8
Excel Modeling BenchmarkFinance & Accounting57.271.7+14.5
LMArena Agent Arena Code Praise vs ComplaintAgents3.031.4+14.2
Senior SWE-BenchCode22.736.8+14.1
BigFinanceBench private holdout (final-answer accuracy)Finance & Accounting40.654.6+14.0
ACCoRDGeneral QA68.181.7+13.6
Biopharma Bench V0.1Health & Medicine57.070.4+13.4
AutomationBench 1.0.6 (provider-reported)Agents54.841.4−13.4
Artificial Analysis Intelligence Index v4.3Indexes39.552.8+13.3
Artificial Analysis Intelligence Index v4.3.2Indexes39.552.7+13.2
SpatialBench-Long V1 Clean RerunsScience26.239.3+13.1
Dual Exam Agent IndexIndexes18.331.3+13.0
Arena Alignment IndexSafety74.887.8+13.0
TxBench Preclinical Pharmacology V1 Clean RerunsScience46.759.3+12.7
LMArena Agent Arena Work Praise vs ComplaintAgents3.728.9+12.6
ScBench-Long V1 Clean RerunsScience14.827.3+12.5
MedXpertQA MM Extended (Mercor held-out)Health & Medicine48.460.8+12.4
Medical Long Context Reasoning (MLCR-AA)Health & Medicine22.835.0+12.2
BullshitBench v2 LeaderboardSafety57.069.0+12.0
MetagenomicsBench V0Science38.049.7+11.7
MetagenomicsBench V1Science38.049.7+11.7
BioMysteryBench (Vals Terminus 2)Science67.879.3+11.5
Vibe Code Bench 1–100Code16.427.6+11.3
Bedside BenchHealth & Medicine82.993.4+10.5
Artificial Analysis Progress Index v1Indexes320350+10.3
Epoch Capabilities Index (ECI)Indexes156169+10.3
BioSecBench-Function V0Science38.748.6+9.9
APEX-AccountingFinance & Accounting48.157.9+9.8
Creative Writing v3 Rubric ScoreChat & Writing14.115.9+9.4
ScBench V1 Clean RerunsScience54.763.9+9.2
Hebbia Financial Services BenchmarkFinance & Accounting63.172.3+9.2
τ³-Banking (Fireworks, banking_knowledge)Agents35.744.8+9.1
Finance Agent v2 - General Quantitative AnalysisFinance & Accounting69.878.8+9.0
τ³-Banking (Mercor run, bm25_grep retrieval)Agents37.846.7+8.9
Longform Creative WritingChat & Writing74.082.8+8.8
Vals Index v2Indexes57.966.6+8.7
Extended NYT Connections 940 (Quadratic v1)Games89.698.1+8.5
SciCode (main problems)Science11.219.6+8.4
Finance Agent v2 - ComparablesFinance & Accounting46.137.8−8.3
BioSecBench-RespiratoryVirus-Surveillance V1 (110 evaluations)Science5.213.3+8.2
LiveBench 2.0Agents78.086.1+8.1
AA-Briefcase EloAgents14241562+8.0
ComplexConstraintsChat & Writing49.757.7+8.0
Ramp Accounting BenchFinance & Accounting40.848.7+7.9
FrontierCode 1.1 Extended - ScoreCode56.664.5+7.9
BioSecBench-Function V1 Clean RerunsScience40.848.6+7.8
LMArena Agent Arena Chat Bash ReliabilityAgents-1.014.5+7.7
LMArena Agent Arena Code Steering BurdenAgents2.317.8+7.7
DTBenchGeneral QA89.997.3+7.5
Artificial Analysis Cyber Index v1Indexes40.833.4−7.4
Artificial Analysis Coding Agent Index v2.0Indexes47.354.7+7.4
MedCodeHealth & Medicine41.248.5+7.3
LMArena Agent Arena Chat Confirmed SuccessAgents4.918.8+6.9
Harvey LAB-AA v1.1 (hallucination-gated)Legal1.78.6+6.9
Epiq LAER CorporateBenchProfessional Work75.382.0+6.7
OncologyBench Strict (25 tasks, October 9 rubric)Health & Medicine14.721.3+6.7
LMArena Agent Arena Work Steering BurdenAgents1.514.5+6.5
BioSecBench-Structure V1 (58 evaluations)Science27.621.3−6.3
SheetBench-50Finance & Accounting58.464.7+6.3
Westworld Finance Diligence BenchFinance & Accounting48.354.5+6.2
LMArena Agent Arena Chat Steering BurdenAgents-0.511.9+6.2
CharXiv (descriptive and reasoning combined)Multimodal90.196.1+6.0
MacroscopeBenchCode72.078.0+6.0
DeepsecBench-AA (Artificial Analysis)Cybersecurity31.236.9+5.7
SWE Atlas - Codebase QnACode53.547.8−5.7
LMArena Agent Arena Chat Praise vs ComplaintAgents3.114.3+5.6
DistressBenchSafety79.874.3−5.5
Financial Coaching BenchFinance & Accounting72.867.5−5.3
Vision Arena OverallMultimodal12631283+5.3
SpatialBench V0 VerifiedScience75.470.1−5.2
CWE-Bench-AA (Artificial Analysis)Cybersecurity68.363.3−5.0
Finance Agent v2.1Finance & Accounting65.460.4−5.0
Harvey LAB: In-House Contracting (held-out)Legal0.35.3+5.0
ProgramBenchCode0.55.5+5.0
HealthBenchHealth & Medicine61.956.9−5.0
Excel Modeling Benchmark v1.1Finance & Accounting67.872.7+4.9
SciCode 1.0.1 (Artificial Analysis)Code51.956.5+4.6
SpatialBench V1 Clean RerunsScience63.267.5+4.4
ARC-AGI-1 Semi-PrivateGeneral QA94.598.5+4.0
LMArena Agent Arena Work Bash ReliabilityAgents0.18.1+4.0
AA-PersonalityBench Expressiveness indexChat & Writing61.958.1−3.8
APEX-1 (Mercor, mean score)Professional Work65.569.3+3.8
LMArena Agent Arena Code Confirmed SuccessAgents6.613.6+3.5
Finance Agent v2 - Earnings AnalysisFinance & Accounting72.969.5−3.4
LMArena Agent Arena Code Bash ReliabilityAgents-0.26.6+3.4
LLM Debate BenchmarkChat & Writing15821606+3.3
APEX-SWECode51.948.6−3.3
MMLU ProGeneral QA88.291.5+3.3
SWE-bench Multimodal (Vals v1, 480 Tasks)Code46.543.3−3.1
DeepResearch Bench II (Mercor 130-task set)Search53.250.1−3.1
DualEntry Accounting AI BenchmarkFinance & Accounting84.281.2−3.0
BioSecBench-Surveillance V1 Clean RerunsScience51.548.5−2.9
EpiBench V1 Clean RerunsScience25.528.3+2.8
LMArena Agent Arena Work Confirmed SuccessAgents5.711.2+2.7
HealthBench (length-adjusted, Artificial Analysis)Health & Medicine60.958.2−2.7
GDPval-AA v2Professional Work16321580−2.5
EpiBenchScience42.144.6+2.5
MedScribeHealth & Medicine85.587.9+2.4
Finance Agent v2 - Market AnalysisFinance & Accounting63.465.5+2.2
Finance Agent v2 - PrecedentsFinance & Accounting28.230.3+2.1
Immigration BenchLegal65.667.7+2.1
AdvancedIFChat & Writing84.082.0−2.0
Legacy-Bench (Fireworks, avg@3)Code57.755.7−2.0
Legal Research BenchLegal41.339.4−1.9
AIME 2025Math98.3100.0+1.7
ITBench-AAAgents46.948.6+1.7
OTIS Mock AIME 2024-2025Math98.3100.0+1.7
SAGEMath47.946.4−1.5
Finance Agent v2 - Financial ModelingFinance & Accounting23.121.6−1.5
Vals Finance IndexIndexes57.155.6−1.4
LMArena Agent Arena Bash RecoveryAgents7.95.5−1.2
EnterpriseBench: CoreCraftProfessional Work41.542.6+1.1
Finance Agent v2 - General Qualitative AnalysisFinance & Accounting67.468.4+1.0
LMArena Agent Arena SteerabilityAgents-3.3-1.4+1.0
Finance Agent v2 - AdjustmentsFinance & Accounting49.248.3−0.9
Tax Agent BenchFinance & Accounting62.563.3+0.9
Nimble Web ContextSearch73.572.6−0.8
SpeciEval (Pooled English Runs)Safety97.496.5−0.8
SpeciEval (Refusal-Aware Retries, Historical)Safety97.496.5−0.8
Finance Agent v2 - All PassFinance & Accounting41.140.3−0.8
NeuroscienceBench V1 (91 evaluations)Science22.022.7+0.7
Terminal-Bench 2.1Code90.689.9−0.7
GDPval-AA v2.1Professional Work16001587−0.6
Harvey LAB held-out set (Fireworks run)Legal5.75.1−0.6
AutomationBench-AA 1.0.6Agents68.968.5−0.4
Finance Agent v2 - Disclosure AnalysisFinance & Accounting61.361.7+0.4
LMArena Text Arena EloChat & Writing14751474−0.2
LMArena Agent Arena Work Tool HallucinationAgents0.50.7+0.1
DeepSWE v1.1Code74.274.1−0.1
LMArena Agent Arena Chat Tool HallucinationAgents0.30.5+0.1
LMArena Agent Arena Code Tool HallucinationAgents0.50.7+0.1
Harvey HLAB — Criteria Pass RateLegal87.587.50.0
GDPval (Mercor 213-task rubric score)Professional Work86.786.70.0
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 - Environmental ESGLegal0.00.00.0
Harvey's Legal Agent Benchmark - Funds Asset ManagementLegal0.00.00.0
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 - Structured Finance SecuritizationLegal0.00.00.0
Harvey's Legal Agent Benchmark - TaxLegal0.00.00.0
Harvey's Legal Agent Benchmark - Trusts Estates Private ClientLegal20.020.00.0
Harvey's Legal Agent Benchmark - White Collar Defense InvestigationsLegal0.00.00.0
SpatialBench-Long V0 VerifiedScience34.934.90.0

DeepSeek V4.1 FlashvsGPT-6 Astranormalised score marginsData Nov 15, 2026Data: Vals AI, Artificial Analysis, benchmarks.bio and 51 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…