Atlas

Compare Models

Gemini 3.7 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 Gemini 3.7 Flash · P(higher) ≈ 100%

Wins 43 of 64 shared benchmarks (20 lost, 1 tied)

Capability vs. relative cost
Gemini 3.7 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.

Gemini 3.7 FlashvsGPT-6.1 Terracapability vs. relative costData Nov 15, 2026Village LLM Atlasatlas.village.dev

Benchmark wins20 – 43 · 64 shared, 1 tied

Comparing matching setups only (12 benchmarks excluded) · suites counted once
← Gemini 3.7 FlashGPT-6.1 Terra →

Gemini 3.7 FlashvsGPT-6.1 Terrawins by categoryData Nov 15, 2026Data: Artificial Analysis, benchmarks.bio, 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 Gemini 3.7 Flash.

FrontierMath Tier 4 (v2) - PrivateMath36.692.7+56.1
Furniture Assembly BenchmarkMultimodal26.779.2+52.5
Terminal-Bench 4.0Code6.150.3+44.2
Mystery Game PuzzlesGames37.074.0+37.0
ALE-BenchCode18762875+35.4
Terminal-Bench 4.0 (Artificial Analysis)Code13.641.9+28.3
LMArena Agent Arena Net Improvement (Steering Burden)Agents-2.212.0+26.8
GBENCH One-shot Coding GScoreGames52.677.6+25.0
ChartographyMultimodal40.463.2+22.8
BioSecBench-Refusal V1 Clean RerunsSafety54.834.2−20.6
FrontierMath Tiers 1-3 (v2) - PrivateMath71.690.9+19.3
Medical Long Context Reasoning (MLCR-AA, October 2026 methodology)Health & Medicine12.230.6+18.3
Humanity's Last Exam DiamondGeneral QA31.247.3+16.1
SpatialBench V1 Clean RerunsScience59.173.3+14.2
CritPt (avg@5)Science14.328.0+13.7
Vibe Code Bench v1.1Code70.484.0+13.6
LiveBench 2.0Agents67.279.7+12.5
LMArena Agent Arena Steering BurdenAgents-7.316.6+12.0
SimpleQA Verified (Epoch anti-abstention)General QA69.257.3−11.9
LLM Chess EloGames11071391+11.7
LMArena Agent Arena Praise vs ComplaintAgents-2.920.0+11.5
SpatialBench-Long V1 Clean RerunsScience29.840.5+10.7
TxBench Antibody Discovery V1 (100 evaluations)Science50.740.7−10.0
TaxCalcBench TY25Finance & Accounting14.024.0+10.0
Artificial Analysis Progress Index v1Indexes317345+9.8
Artificial Analysis Intelligence Index v4.3.2Indexes39.148.1+9.0
Legal Research BenchLegal34.626.0−8.7
AA-Briefcase v1.1 EloAgents11151265+8.3
NoCoT-Bench NCKI (kspine v3)General QA121110−7.9
ScBench V1 Clean RerunsScience50.458.3+7.9
Epoch Capabilities Index (ECI)Indexes158168+7.8
TxBench Oligo Discovery V1 (113 evaluations)Science37.830.4−7.4
Humanity's Last Exam (Text-Only)General QA39.046.2+7.3
MetagenomicsBench V1Science31.738.7+7.0
LMArena Agent Arena Confirmed SuccessAgents-3.59.7+6.6
APEX-Accounting 1.1Finance & Accounting57.263.7+6.5
BuseyBench SVGMultimodal6.67.3+6.4
BioSecBench-Function V1 Clean RerunsScience33.739.0+5.4
GDP.pdf (Artificial Analysis, All-pass)Professional Work23.628.6+5.0
SciCode 1.0.1 (Artificial Analysis)Code59.855.1−4.7
LMArena Agent Arena Bash ReliabilityAgents6.213.4+3.6
VariantBench V1 Clean RerunsScience31.434.8+3.4
DistressBenchSafety69.065.9−3.1
Toloka Arena (v3)Agents58.055.0−3.0
APEX-1 (Mercor, mean score)Professional Work67.064.2−2.8
OTIS Mock AIME 2024-2025Math97.2100.0+2.8
Tax Agent BenchFinance & Accounting57.755.0−2.6
NoCoT-Bench NCRI 15.2General QA126130+2.4
GDPval-AA v2.1Professional Work14041353−2.2
AutomationBench-AA 1.0.6Agents62.060.2−1.9
AA-Omniscience IndexIndexes26.529.8+1.6
ChessBench CoherenceGames83.284.8+1.6
BioSecBench-Surveillance V1 Clean RerunsScience41.539.9−1.5
OncologyBench Strict (25 tasks, October 9 rubric)Health & Medicine14.716.0+1.3
EpiBench V1 Clean RerunsScience22.023.3+1.3
ChessBench AccuracyGames79.278.3−1.0
ChessBench EloGames12651251−0.8
Design Arena Overall Code CategoriesCode13041300−0.7
TxBench Preclinical Pharmacology V1 Clean RerunsScience55.355.7+0.3
AA-LCR v1.1Search83.082.7−0.3
MMMU ProMultimodal84.984.7−0.2
LMArena Agent Arena Tool HallucinationAgents0.71.0+0.2
APEX-Agents 1.1Professional Work79.479.3−0.1
ScBench-Long V1 Clean RerunsScience18.818.80.0

Gemini 3.7 FlashvsGPT-6.1 Terranormalised score marginsData Nov 15, 2026Data: Artificial Analysis, benchmarks.bio, 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…