Atlas

Compare Models

DeepSeek V4.1 Flash vs GPT-6.1 Terra

GPT-6.1 Terra leads by 5.9 capability points (≈95% interval +3.7 to +8.1): a clear lead over DeepSeek V4.1 Flash · P(higher) ≈ 100%

Wins 52 of 71 shared benchmarks (19 lost)

Capability vs. relative cost
DeepSeek V4.1 FlashGPT-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 FlashvsGPT-6.1 Terracapability vs. relative costData Nov 15, 2026Village LLM Atlasatlas.village.dev

Benchmark wins19 – 52 · 71 shared

Comparing matching setups only (6 benchmarks excluded) · suites counted once
← DeepSeek V4.1 FlashGPT-6.1 Terra →

DeepSeek V4.1 FlashvsGPT-6.1 Terrawins by categoryData Nov 15, 2026Data: benchmarks.bio, Artificial Analysis, Epoch AI and 20 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.

FrontierMath Tier 4 (v2) - PrivateMath26.892.7+65.9
ChartographyMultimodal11.663.2+51.6
Terminal-Bench-Science 0.1Science9.057.6+48.6
Furniture Assembly BenchmarkMultimodal34.279.2+45.0
ALE-BenchCode18292875+37.1
BuseyBench SVGMultimodal4.17.3+31.9
Mystery Game PuzzlesGames43.074.0+31.0
GBENCH One-shot Coding GScoreGames48.077.6+29.6
APEX-Agents 1.1Professional Work50.579.3+28.8
Terminal-Bench 4.0 (Artificial Analysis)Code26.855.6+28.8
Humanity's Last Exam DiamondGeneral QA19.147.3+28.2
RareDiseaseBench V1 (35 evaluations)Science0.025.7+25.7
FrontierMath Tiers 1-3 (v2) - PrivateMath67.490.9+23.5
TissueMicroscopyBench V1 (49 evaluations)Science2.725.9+23.1
BioSecBench-Refusal V1 Clean RerunsSafety11.334.2+22.9
WeirdML v3Science5.828.1+22.3
TaxCalcBench TY25Finance & Accounting4.024.0+20.0
Terminal-Bench 4.0Code31.250.3+19.1
DAXBench v2Code75.593.6+18.1
AA-Omniscience IndexIndexes-5.329.8+17.5
GameBench 2Games68.485.8+17.4
GDP.pdf (Artificial Analysis, All-pass)Professional Work12.828.6+15.8
Legal Research BenchLegal41.326.0−15.4
HealthBench Professional (Length-Adjusted)Health & Medicine47.862.7+14.9
SimpleQA Verified (Epoch anti-abstention)General QA42.957.3+14.4
SpatialBench-Long V1 Clean RerunsScience26.240.5+14.3
DistressBenchSafety79.865.9−13.9
CritPt (avg@5)Science14.328.0+13.7
APEX-Accounting 1.1Finance & Accounting50.163.7+13.6
LLM Chess EloGames10871391+12.5
AA-LCR v1.1Search55.367.7+12.3
MacroscopeBenchCode70.582.3+11.8
BioSecBench-Surveillance V1 Clean RerunsScience51.539.9−11.5
MMMU ProMultimodal65.676.0+10.4
Humanity's Last Exam (Text-Only)General QA10.821.1+10.3
SpatialBench V1 Clean RerunsScience63.273.3+10.1
Finance Agent v2.1Finance & Accounting65.456.0−9.5
Epoch Capabilities Index (ECI)Indexes156168+9.4
TxBench Preclinical Pharmacology V1 Clean RerunsScience46.755.7+9.0
Artificial Analysis Progress Index v1Indexes320345+8.8
AutomationBench-AA 1.0.6Agents68.960.2−8.7
Artificial Analysis Intelligence Index v4.3.2Indexes39.548.1+8.7
TxBench-Antibody-Development V1 (60 evaluations)Science14.422.8+8.3
FrontierCode 1.1 Main (Artificial Analysis)Code44.036.0−8.0
Tax Agent BenchFinance & Accounting62.555.0−7.4
Medical Long Context Reasoning (MLCR-AA, October 2026 methodology)Health & Medicine23.930.6+6.7
BioSecBench-RespiratoryVirus-Surveillance V1 (110 evaluations)Science5.211.2+6.1
DualEntry Accounting AI BenchmarkFinance & Accounting84.278.2−6.0
Excel Modeling Benchmark v1.1Finance & Accounting67.873.3+5.6
GDPval-AA v2.1Professional Work16001476−5.5
TxBench Oligo Discovery V1 (113 evaluations)Science34.830.4−4.4
ScBench-Long V1 Clean RerunsScience14.818.8+4.0
Vals Finance IndexIndexes57.153.3−3.8
VariantBench V1 Clean RerunsScience31.134.8+3.7
ScBench V1 Clean RerunsScience54.758.3+3.6
SciCode 1.0.1 (Artificial Analysis)Code51.955.1+3.2
Design Arena Overall Code CategoriesCode13161300−2.7
EpiBench V1 Clean RerunsScience25.523.3−2.2
BioSecBench-Function V1 Clean RerunsScience40.839.0−1.8
BioSecBench-Structure V1 (58 evaluations)Science27.625.9−1.7
OTIS Mock AIME 2024-2025Math98.3100.0+1.7
LiveBench 2.0Agents78.079.7+1.6
OncologyBench Strict (25 tasks, October 9 rubric)Health & Medicine14.716.0+1.3
APEX-1 (Mercor, mean score)Professional Work65.564.2−1.3
TxBench Antibody Discovery V1 (100 evaluations)Science39.740.7+1.0
Artificial Analysis Coding Agent Index v2.0Indexes47.348.1+0.8
AA-Briefcase v1.1 EloAgents14151401−0.7
Vibe Code Bench v1.1Code84.784.0−0.7
τ³-Banking (Mercor run, bm25_grep retrieval)Agents37.837.1−0.7
MetagenomicsBench V1Science38.038.7+0.7
NeuroscienceBench V1 (91 evaluations)Science22.022.3+0.4

DeepSeek V4.1 FlashvsGPT-6.1 Terranormalised score marginsData Nov 15, 2026Data: benchmarks.bio, Artificial Analysis, Epoch AI and 20 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…