Atlas

Compare Models

DeepSeek V4.1 Flash 1017 vs GPT-6.1 Sol

GPT-6.1 Sol leads by 5.3 capability points (≈95% interval +2.8 to +7.9): a clear lead over DeepSeek V4.1 Flash 1017 · P(higher) ≈ 100%

Wins 97 of 123 shared benchmarks (26 lost)

Capability vs. relative cost
DeepSeek V4.1 Flash 1017GPT-6.1 SolClosed 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 Flash 1017vsGPT-6.1 Solcapability vs. relative costData Nov 15, 2026Village LLM Atlasatlas.village.dev

Benchmark wins26 – 97 · 123 shared

Comparing matching setups only (52 benchmarks excluded) · suites counted once
← DeepSeek V4.1 Flash 1017GPT-6.1 Sol →

DeepSeek V4.1 Flash 1017vsGPT-6.1 Solwins by categoryData Nov 15, 2026Data: Artificial Analysis, Vals AI, Surge AI and 34 othersVillage LLM Atlasatlas.village.dev

Margins

Each benchmark scaled 0–100% (its declared range, else the field’s range); Δ is GPT-6.1 Sol minus DeepSeek V4.1 Flash 1017.

ARC-AGI-3 (Semi-Private)General QA5.896.2+90.4
Concept Synth INDUCTION Challenge100 Round 1General QA21.090.0+69.0
Concept Synth INDUCTION Challenge64 Round 1General QA32.895.3+62.5
HardGeoBench Avg DistanceMultimodal32011239+60.5
FrontierMath Tier 4 (v2) - PrivateMath46.3100.0+53.7
Terminal-Bench-Science 0.1Science12.958.1+45.2
AA-PersonalityBench Informativeness indexChat & Writing71.427.7−44.2
RiemannBench (November 2026 rescoring)Math38.479.2+40.8
AA-PersonalityBench Conviction indexChat & Writing94.354.3−40.4
LisanBench Difficulty-WeightedGames77334004−39.9
AA-PersonalityBench Warmth indexChat & Writing27.765.7+38.5
LisanBench Path LengthGames211078736−38.4
Humanity's Last Exam DiamondGeneral QA20.858.8+38.0
ATLAS (Exa, 200-task subset)Search31.969.5+37.6
HardGeoBench Country AccuracyMultimodal15.052.5+37.5
NoCoT-Bench NCRI 15.2General QA81.6149+37.0
HardGeoBench <=100miMultimodal7.543.8+36.3
BuseyBench SVGMultimodal4.38.0+36.0
ALE-BenchCode20122990+34.6
SimpleQA Verified (Epoch anti-abstention)General QA42.573.9+31.4
HardGeoBench <=10miMultimodal3.835.0+31.3
Mystery Game PuzzlesGames49.080.0+31.0
CritPt v2 (agentic, avg@5)Science52.382.0+29.7
WeirdML v3Science9.238.3+29.1
Artificial Analysis Intelligence Index v5.0Indexes28.554.5+26.0
DAYJOB: LawLegal6.231.8+25.6
GBENCH One-shot Coding GScoreGames53.879.0+25.2
AA-PersonalityBench Imagination indexChat & Writing37.261.9+25.0
AA-PersonalityBench Slop indexChat & Writing39.114.3−25.0
Terminal-Bench 4.0Code31.855.1+23.2
FrontierMath Tiers 1-3 (v2) - PrivateMath71.293.7+22.5
Terminal-Bench 4.0 (Artificial Analysis)Code33.856.1+22.2
GameBench 2Games69.992.0+22.1
Artificial Analysis Intelligence Index v5.0.1Indexes32.654.6+22.0
Chess Puzzles (task version 2.0)Games46.068.0+22.0
Artificial Analysis Engineering IndexIndexes37.659.4+21.8
DAYJOB: Finance (October 2026 rescoring)Finance & Accounting12.533.5+21.0
Code MigrationCode44.265.1+20.9
LMArena Agent Arena Net Improvement (Steering Burden)Agents2.012.5+19.7
NoCoT-Bench NCKI (kspine v3)General QA102130+19.4
DAYJOB: Healthcare (October 2026 rescoring)Health & Medicine6.025.2+19.2
VIALS-AA (pass^3)Multimodal40.459.0+18.6
AA-Omniscience IndexIndexes4.741.5+18.4
Artificial Analysis Economics IndexIndexes38.556.3+17.8
DAYJOB: AccountingFinance & Accounting7.625.2+17.6
HoneyBench v0.2 (hack rate)Safety76.459.5+16.9
Humanity's Sixth SenseMultimodal30.146.6+16.5
ARC-AGI-2General QA78.194.2+16.1
Artificial Analysis Legal IndexIndexes37.753.7+15.9
WARC-Bench (192-task test split)Agents70.586.3+15.8
DAYJOB: Data ScienceProfessional Work3.318.7+15.4
DAXBench v2Code77.792.6+14.9
GDP.pdf (Artificial Analysis, All-pass)Professional Work17.232.0+14.8
Artificial Analysis Finance & Accounting IndexIndexes42.256.2+14.1
Hebbia Financial Services BenchmarkFinance & Accounting60.373.5+13.2
LMArena Agent Arena Praise vs ComplaintAgents1.227.5+13.2
Tax Agent BenchFinance & Accounting74.962.3−12.6
HealthBench Professional (Length-Adjusted)Health & Medicine51.764.2+12.5
FrontierCode 2.0 (Artificial Analysis)Code28.540.6+12.2
Medical Long Context Reasoning (MLCR-AA, October 2026 methodology)Health & Medicine26.738.3+11.7
VIALS-AA (average of 3)Multimodal53.264.8+11.6
Design Arena Overall Code CategoriesCode12891356+11.1
Artificial Analysis Healthcare & Medical IndexIndexes41.352.0+10.7
Wiz Cyber Model Arena v3.9Cybersecurity60.370.7+10.4
Harvey LAB (Mercor run, undisclosed ~213-task cohort)Legal18.88.8−10.0
Code Migration v1.1Code67.577.3+9.8
DeepsecBench — Revalidated V1 (233 issues)Code30.039.5+9.5
MKA (Artificial Analysis)Professional Work38.547.7+9.2
Excel Modeling BenchmarkFinance & Accounting61.870.8+9.0
Artificial Analysis Progress Index v1Indexes324348+8.7
Humanity's Last ExamGeneral QA42.651.3+8.7
Artificial Analysis Strategy & Ops IndexIndexes49.257.6+8.4
Bedside BenchHealth & Medicine85.793.9+8.3
EnterpriseBench: CoreCraftProfessional Work58.550.3−8.2
HLAB-Vals-1.1Legal14.26.3−7.9
CritPt (avg@5)Science24.031.7+7.7
Epoch Capabilities Index (ECI)Indexes159169+7.7
AA-PersonalityBench Organization indexChat & Writing88.681.0−7.7
Access to Justice BenchmarkLegal73.165.8−7.3
Epiq LAER CorporateBenchProfessional Work75.282.2+7.0
Artificial Analysis Intelligence Index v4.3.2Indexes45.151.8+6.7
Legal Research BenchLegal45.238.5−6.7
LiveBench 2.0Agents79.285.4+6.2
APEX-Agents 1.1Professional Work79.473.2−6.2
AA-PersonalityBench Transparency indexChat & Writing86.792.4+5.8
FrontierCode 1.1 Main (Artificial Analysis)Code43.749.1+5.5
MMMU ProMultimodal79.785.1+5.4
Vals Finance IndexIndexes60.955.8−5.1
APEX-Accounting 1.1Finance & Accounting59.364.3+5.0
Finance Agent v2Finance & Accounting57.052.0−4.9
o11y-bench 1.1 Pass^3Agents79.784.4+4.7
LMArena Agent Arena Confirmed SuccessAgents2.311.7+4.7
LMArena Agent Arena Steering BurdenAgents4.013.4+4.7
HANDBOOK.md Agents v1.1Professional Work24.028.6+4.6
Vals Index v2.1Indexes56.661.2+4.6
Visa Agentic Commerce BenchmarkAgents57.862.2+4.4
VIALS-AA (pass@3)Multimodal66.570.8+4.3
ARC-AGI-1 Semi-PrivateGeneral QA94.598.5+4.0
Harvey's Legal Agent BenchmarkLegal9.25.4−3.7
Vals Legal IndexIndexes71.968.7−3.1
Ruby LLM Benchmarks — Program FixerCode72.075.0+3.1
GDPxProfessional Work5.48.4+3.0
AA-Briefcase v1.1 EloAgents14161469+2.9
DAYJOB: TaxFinance & Accounting1.24.0+2.8
Vals Index v2.2Indexes61.363.9+2.7
PI OfficeQA Pro v2Professional Work78.881.1+2.3
APEX-SWECode49.146.9−2.2
τ³-Banking (Fireworks, banking_knowledge)Agents46.844.7−2.1
Bug Hunt BenchCode40.242.2+2.0
AA-PersonalityBench Expressiveness indexChat & Writing52.450.5−1.9
AutomationBench-AA 1.0.6Agents68.566.6−1.9
Excel Modeling Benchmark v1.1Finance & Accounting69.271.1+1.9
τ³-Banking (Mercor run, bm25_grep retrieval)Agents41.643.0+1.4
AA-LCR v1.1Search82.784.0+1.3
Legacy-Bench (Fireworks, avg@3)Code56.355.0−1.3
LMArena Agent Arena Bash ReliabilityAgents2.45.0+1.3
Vibe Code Bench v1.1Code87.788.9+1.2
Finance Agent v2.1Finance & Accounting62.961.8−1.1
DualEntry Accounting AI BenchmarkFinance & Accounting82.281.2−1.0
OTIS Mock AIME 2024-2025Math99.4100.0+0.6
LMArena Agent Arena Tool HallucinationAgents0.31.0+0.4
DeepSWE v1.1 (Fireworks, avg@3)Code69.669.3−0.3
SciCode 1.0.1 (Artificial Analysis)Code55.955.8−0.1

DeepSeek V4.1 Flash 1017vsGPT-6.1 Solnormalised score marginsData Nov 15, 2026Data: Artificial Analysis, Vals AI, Surge AI and 34 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…