Atlas

Compare Models

DeepSeek V4.1 Flash 1017 vs GPT-6.1 Terra

GPT-6.1 Terra leads by 4.4 capability points (≈95% interval +1.8 to +6.9): a clear lead over DeepSeek V4.1 Flash 1017 · P(higher) ≈ 100%

Wins 57 of 77 shared benchmarks (19 lost, 1 tied)

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

Benchmark wins19 – 57 · 77 shared, 1 tied

Comparing matching setups only (18 benchmarks excluded) · suites counted once
← DeepSeek V4.1 Flash 1017GPT-6.1 Terra →

DeepSeek V4.1 Flash 1017vsGPT-6.1 Terrawins by categoryData Nov 15, 2026Data: Artificial Analysis, Vals AI, Epoch AI and 18 othersVillage LLM Atlasatlas.village.dev

Margins

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

HardGeoBench Avg DistanceMultimodal32011091+65.1
FrontierMath Tier 4 (v2) - PrivateMath46.392.7+46.3
Terminal-Bench-Science 0.1Science12.957.6+44.8
HardGeoBench Country AccuracyMultimodal15.058.8+43.8
HardGeoBench <=100miMultimodal7.546.3+38.8
HardGeoBench <=10miMultimodal3.835.0+31.3
CritPt v2 (agentic, avg@5)Science50.681.4+30.9
ALE-BenchCode20122875+30.6
BuseyBench SVGMultimodal4.37.3+29.3
NoCoT-Bench NCRI 15.2General QA81.6130+26.8
Humanity's Last Exam DiamondGeneral QA20.847.3+26.5
Mystery Game PuzzlesGames49.074.0+25.0
GBENCH One-shot Coding GScoreGames53.877.6+23.8
Artificial Analysis Intelligence Index v5.0Indexes28.551.0+22.4
Terminal-Bench 4.0 (Artificial Analysis)Code33.855.6+21.7
VIALS-AA (pass^3)Multimodal40.460.2+19.9
Tax Agent BenchFinance & Accounting74.955.0−19.9
FrontierMath Tiers 1-3 (v2) - PrivateMath71.290.9+19.6
Legal Research BenchLegal45.226.0−19.2
WeirdML v3Science9.228.1+18.9
LMArena Agent Arena Net Improvement (Steering Burden)Agents2.012.0+18.9
Artificial Analysis Intelligence Index v5.0.1Indexes32.651.3+18.7
Terminal-Bench 4.0Code31.849.5+17.7
Artificial Analysis Engineering IndexIndexes37.654.7+17.1
DAXBench v2Code77.793.6+16.0
GameBench 2Games69.985.8+15.9
SimpleQA Verified (Epoch anti-abstention)General QA42.557.3+14.8
Harvey LAB (Mercor run, undisclosed ~213-task cohort)Legal18.84.5−14.3
FrontierCode 2.0 (Artificial Analysis)Code28.542.3+13.8
Vals Legal IndexIndexes71.958.8−13.1
VIALS-AA (average of 3)Multimodal53.266.0+12.8
AA-Omniscience IndexIndexes4.729.8+12.5
HLAB-Vals-1.1Legal14.21.7−12.5
GDP.pdf (Artificial Analysis, All-pass)Professional Work17.228.6+11.4
HealthBench Professional (Length-Adjusted)Health & Medicine51.762.7+11.0
Artificial Analysis Economics IndexIndexes38.548.5+10.0
LMArena Agent Arena Praise vs ComplaintAgents1.220.0+9.4
Artificial Analysis Legal IndexIndexes37.746.4+8.7
AutomationBench-AA 1.0.6Agents68.560.2−8.3
AA-Briefcase v1.1 EloAgents14161265−8.3
FrontierCode 1.1 Main (Artificial Analysis)Code43.736.0−7.6
Vals Finance IndexIndexes60.953.3−7.6
Artificial Analysis Progress Index v1Indexes324345+7.6
Epoch Capabilities Index (ECI)Indexes159168+7.1
Finance Agent v2.1Finance & Accounting62.956.0−7.0
Artificial Analysis Finance & Accounting IndexIndexes42.249.0+6.8
Visa Agentic Commerce BenchmarkAgents57.851.1−6.7
LMArena Agent Arena Steering BurdenAgents4.016.6+6.3
NoCoT-Bench NCKI (kspine v3)General QA102110+5.6
LMArena Agent Arena Bash ReliabilityAgents2.413.4+5.5
MKA (Artificial Analysis)Professional Work38.543.8+5.2
MMMU ProMultimodal79.784.7+5.0
VIALS-AA (pass@3)Multimodal66.571.4+5.0
τ³-Banking (Mercor run, bm25_grep retrieval)Agents41.637.1−4.5
APEX-Accounting 1.1Finance & Accounting59.363.7+4.4
Excel Modeling Benchmark v1.1Finance & Accounting69.273.3+4.2
APEX-1 (Mercor, mean score)Professional Work68.364.2−4.1
CritPt (avg@5)Science24.028.0+4.0
DualEntry Accounting AI BenchmarkFinance & Accounting82.278.2−4.0
Medical Long Context Reasoning (MLCR-AA, October 2026 methodology)Health & Medicine26.730.6+3.9
Vibe Code Bench v1.1Code87.784.0−3.7
Code Migration v1.1 (Vals Index v2.2 subset)Code68.572.2+3.7
LMArena Agent Arena Confirmed SuccessAgents2.39.7+3.7
Artificial Analysis Healthcare & Medical IndexIndexes41.344.7+3.4
GDPxProfessional Work5.48.5+3.1
Artificial Analysis Intelligence Index v4.3.2Indexes45.148.1+3.0
Humanity's Last Exam (Text-Only)General QA43.346.2+2.9
Design Arena Overall Code CategoriesCode12891300+1.8
Bug Hunt BenchCode40.238.7−1.5
SciCode 1.0.1 (Artificial Analysis)Code55.955.1−0.8
Artificial Analysis Strategy & Ops IndexIndexes49.250.0+0.8
Vals Index v2.2Indexes61.360.5−0.7
OTIS Mock AIME 2024-2025Math99.4100.0+0.6
LiveBench 2.0Agents79.279.7+0.5
LMArena Agent Arena Tool HallucinationAgents0.31.0+0.4
APEX-Agents 1.1Professional Work79.479.3−0.1
AA-LCR v1.1Search82.782.70.0

DeepSeek V4.1 Flash 1017vsGPT-6.1 Terranormalised score marginsData Nov 15, 2026Data: Artificial Analysis, Vals AI, Epoch AI and 18 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…