Atlas

Compare Models

Gemini 4 Argon vs GPT-6.1 Terra

Gemini 4 Argon leads by 0.6 capability points (≈95% interval −1.9 to +3.1): statistically tied · P(higher) ≈ 69%

Wins 73 of 105 shared benchmarks (31 lost, 1 tied)

Capability vs. relative cost
Gemini 4 ArgonGPT-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.

Gemini 4 ArgonvsGPT-6.1 Terracapability vs. relative costData Nov 15, 2026Village LLM Atlasatlas.village.dev

Benchmark wins73 – 31 · 105 shared, 1 tied

Comparing matching setups only (17 benchmarks excluded) · suites counted once
← Gemini 4 ArgonGPT-6.1 Terra →

Gemini 4 ArgonvsGPT-6.1 Terrawins by categoryData Nov 15, 2026Data: Artificial Analysis, benchmarks.bio, Epoch AI and 28 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 Gemini 4 Argon.

Visa Agentic Commerce BenchmarkAgents82.251.1−31.1
Legal Research BenchLegal54.826.0−28.8
HLAB-Vals-1.1Legal27.11.7−25.4
Tax Agent BenchFinance & Accounting79.055.0−24.0
Terminal-Bench-Science 0.1Science48.624.8−23.8
TxBench Antibody Discovery V1 (100 evaluations)Science63.740.7−23.0
Harvey LAB (Mercor run, undisclosed ~213-task cohort)Legal27.44.5−22.9
Terminal-Bench 4.0 (Artificial Analysis)Code55.132.3−22.7
τ³-Banking (Mercor run, bm25_grep retrieval)Agents58.137.1−21.0
TxBench Oligo Discovery V1 (113 evaluations)Science51.030.4−20.6
BioSecBench-Surveillance V1 Clean RerunsScience59.839.9−19.9
SimpleQA Verified (Epoch anti-abstention)General QA76.057.3−18.7
RareDiseaseBench V1 (35 evaluations)Science43.825.7−18.1
Vals Legal IndexIndexes76.858.8−18.0
Toloka Arena (v3)Agents73.055.0−18.0
FrontierMemory v1.1Agents64.446.7−17.8
AutomationBench-AA 1.0.6Agents77.560.2−17.3
OncologyBench Strict (25 tasks, October 9 rubric)Health & Medicine33.316.0−17.3
Medical Long Context Reasoning (MLCR-AA, October 2026 methodology)Health & Medicine13.930.6+16.7
VariantBench V1 Clean RerunsScience50.334.8−15.5
FrontierMath Tier 4 (v2) - PrivateMath78.092.7+14.6
BioSecBench-Refusal V1 Clean RerunsSafety48.734.2−14.5
ChessBench EloGames14851251−13.9
EBR-bench v4 (card ban, single agent)Games64.878.1+13.3
NoCoT-Bench NCKI (kspine v3)General QA128110−12.9
DualEntry Accounting AI BenchmarkFinance & Accounting91.178.2−12.9
Vals Finance IndexIndexes66.053.3−12.7
HardGeoBench <=100miMultimodal58.846.3−12.5
SWE Atlas v1.1 - OverallCode70.157.7−12.4
GDPval-AA v2.1Professional Work16341353−12.3
Blueprint-Bench 2Multimodal0.50.4−12.3
AA-Briefcase v1.1 EloAgents14851265−12.0
SpatialBench-Long V1 Clean RerunsScience28.640.5+11.9
FrontierCode 2.0 (Artificial Analysis)Code25.737.5+11.8
Finance Agent v2.1Finance & Accounting67.756.0−11.7
NoCoT-Bench NCRI 15.2General QA151130−11.4
BioSecBench-Structure V1 (58 evaluations)Science14.925.9+10.9
Humanity's Last Exam (Text-Only)General QA57.146.2−10.8
HardGeoBench Avg DistanceMultimodal7411091−10.8
HardGeoBench <=10miMultimodal45.035.0−10.0
Mystery Game PuzzlesGames64.074.0+10.0
GBENCH One-shot Coding GScoreGames67.977.6+9.7
ScBench-Long V1 Clean RerunsScience28.418.8−9.7
MacroscopeBenchCode72.882.3+9.5
LMArena Agent Arena Bash ReliabilityAgents-5.413.4+9.4
TxBench Preclinical Pharmacology V1 Clean RerunsScience64.755.7−9.0
LLM Chess EloGames16091391−9.0
ChartographyMultimodal71.663.2−8.4
TxBench-Antibody-Development V1 (60 evaluations)Science31.122.8−8.3
APEX-Agents 1.1Professional Work87.479.3−8.1
APEX-Accounting 1.1Finance & Accounting71.563.7−7.8
ChessBench CoherenceGames92.684.8−7.8
MKA (Artificial Analysis)Professional Work51.343.8−7.5
Terminal-Bench 4.0Code57.450.3−7.1
SciCode 1.0.1 (Artificial Analysis)Code61.855.1−6.7
Furniture Assembly BenchmarkMultimodal72.579.2+6.7
BioSecBench-RespiratoryVirus-Surveillance V1 (110 evaluations)Science17.611.2−6.4
FrontierCode 1.1 Main (Artificial Analysis)Code43.537.2−6.3
AA-Omniscience IndexIndexes42.429.8−6.3
ChessBenchGames70.064.0−6.0
HiL-Bench (Human-in-Loop Benchmark)Agents62.068.0+6.0
EpiBench V1 Clean RerunsScience28.923.3−5.7
BioSecBench-Function V1 Clean RerunsScience44.439.0−5.4
DAXBench v2Code88.393.6+5.3
HardGeoBench Country AccuracyMultimodal63.858.8−5.0
FrontierMath Tiers 1-3 (v2) - PrivateMath86.090.9+4.9
Humanity's Last Exam DiamondGeneral QA42.547.3+4.8
MultiChallengeChat & Writing78.774.2−4.5
Artificial Analysis Intelligence Index v4.3.2Indexes52.648.1−4.5
LiveBench 2.0Agents84.179.7−4.4
SpatialBench V1 Clean RerunsScience69.073.3+4.3
Design Arena Overall Code CategoriesCode13261300−4.3
TissueMicroscopyBench V1 (49 evaluations)Science21.825.9+4.1
CritPt v2 (agentic, avg@5)Science72.976.9+4.0
TaxCalcBench TY25 (web search)Finance & Accounting74.070.0−4.0
ChessBench AccuracyGames81.878.3−3.8
MetagenomicsBench V1Science42.338.7−3.7
MMMU ProMultimodal88.384.7−3.6
GameBench 2Games89.285.8−3.3
NeuroscienceBench V1 (91 evaluations)Science19.122.3+3.2
AA-LCR v1.1Search79.782.7+3.0
GDP.pdf (Artificial Analysis, All-pass)Professional Work25.628.6+3.0
BuseyBench SVGMultimodal7.67.3−2.8
DistressBenchSafety68.665.9−2.7
ALE-BenchCode28052875+2.5
LMArena Agent Arena Net Improvement (Steering Burden)Agents10.712.0+2.4
Code Migration v1.1 (Vals Index v2.2 subset)Code74.472.2−2.2
GPQA DiamondGeneral QA96.494.3−2.1
LMArena Agent Arena Praise vs ComplaintAgents24.120.0−2.0
LMArena Agent Arena Confirmed SuccessAgents13.29.7−1.8
LMArena Agent Arena Steering BurdenAgents13.216.6+1.7
Excel Modeling Benchmark v1.1Finance & Accounting74.873.3−1.5
Epoch Capabilities Index (ECI)Indexes167168+1.3
VIALS-AA (pass@3)Multimodal72.771.4−1.2
Artificial Analysis Progress Index v1Indexes342345+1.2
GDPxProfessional Work7.38.5+1.1
Artificial Analysis Intelligence Index v5.0.1Indexes50.451.3+0.9
ScBench V1 Clean RerunsScience59.158.3−0.9
CritPt (avg@5)Science27.128.0+0.9
VIALS-AA (pass^3)Multimodal59.660.2+0.6
Prophet Arena Arena Score (Agentic, OpenCode)Forecasting86.085.7−0.4
LMArena Agent Arena Tool HallucinationAgents0.51.0+0.2
VIALS-AA (average of 3)Multimodal66.366.0−0.2
Prophet Arena Arena Score (Fixed Context)Forecasting85.385.2−0.1
OTIS Mock AIME 2024-2025Math100.0100.00.0

Gemini 4 ArgonvsGPT-6.1 Terranormalised score marginsData Nov 15, 2026Data: Artificial Analysis, benchmarks.bio, Epoch AI and 28 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…