Atlas

Compare Models

Gemini 3.8 Flash vs GPT-6.1 Terra

GPT-6.1 Terra leads by 4.6 capability points (≈95% interval +2.3 to +7.0): a clear lead over Gemini 3.8 Flash · P(higher) ≈ 100%

Wins 69 of 100 shared benchmarks (31 lost)

Capability vs. relative cost
Gemini 3.8 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 3.8 FlashvsGPT-6.1 Terracapability vs. relative costData Nov 15, 2026Village LLM Atlasatlas.village.dev

Benchmark wins31 – 69 · 100 shared

Comparing matching setups only (20 benchmarks excluded) · suites counted once
← Gemini 3.8 FlashGPT-6.1 Terra →

Gemini 3.8 FlashvsGPT-6.1 Terrawins by categoryData Nov 15, 2026Data: Artificial Analysis, benchmarks.bio, Vals AI and 24 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 3.8 Flash.

FrontierMath Tier 4 (v2) - PrivateMath22.092.7+70.7
Furniture Assembly BenchmarkMultimodal31.779.2+47.5
Code Migration v1.1 (Vals Index v2.2 subset)Code38.272.2+34.0
Terminal-Bench 4.0Code19.749.5+29.8
Mystery Game PuzzlesGames47.074.0+27.0
HiL-Bench (Human-in-Loop Benchmark)Agents41.568.0+26.5
ALE-BenchCode21822875+24.6
Vending-Bench 2Games50949107+24.3
ScBench V1 Clean RerunsScience34.058.3+24.3
Terminal-Bench-Science 0.1Science10.033.3+23.3
RareDiseaseBench V1 (35 evaluations)Science2.925.7+22.9
FrontierMath Tiers 1-3 (v2) - PrivateMath68.490.9+22.5
TissueMicroscopyBench V1 (49 evaluations)Science3.425.9+22.5
ChartographyMultimodal40.963.2+22.3
Terminal-Bench 4.0 (Artificial Analysis)Code19.741.9+22.2
GameBench 2Games64.085.8+21.8
Artificial Analysis Intelligence Index v5.0.1Indexes30.851.3+20.5
Artificial Analysis Intelligence Index v5.0Indexes30.551.0+20.4
VIALS-AA (pass^3)Multimodal41.060.2+19.3
GBENCH One-shot Coding GScoreGames58.677.6+18.9
FrontierCode 2.0 (Artificial Analysis)Code18.737.5+18.7
LMArena Agent Arena Net Improvement (Steering Burden)Agents2.312.0+18.3
Humanity's Last Exam DiamondGeneral QA29.147.3+18.2
TaxCalcBench TY25Finance & Accounting6.024.0+18.0
ChessBench EloGames15271251−16.4
Medical Long Context Reasoning (MLCR-AA, October 2026 methodology)Health & Medicine15.630.6+15.0
VIALS-AA (average of 3)Multimodal52.066.0+14.1
SpatialBench V1 Clean RerunsScience59.473.3+13.9
BioSecBench-Refusal V1 Clean RerunsSafety47.434.2−13.2
Legal Research BenchLegal38.926.0−13.0
SimpleQA Verified (Epoch anti-abstention)General QA69.757.3−12.4
LiveBench 2.0Agents67.379.7+12.3
MetagenomicsBench V1Science26.738.7+12.0
SpatialBench-Long V1 Clean RerunsScience28.640.5+11.9
Tax Agent BenchFinance & Accounting66.855.0−11.7
CritPt v2 (agentic, avg@5)Science65.176.9+11.7
τ³-Banking (Mercor run, bm25_grep retrieval)Agents48.837.1−11.7
Vals Legal IndexIndexes69.858.8−11.0
MacroscopeBenchCode70.781.2+10.5
HLAB-Vals-1.1Legal12.11.7−10.4
Finance Agent v2.1Finance & Accounting65.856.0−9.8
ChessBench CoherenceGames94.684.8−9.8
CritPt (avg@5)Science18.328.0+9.7
ChessBenchGames73.364.0−9.3
LMArena Agent Arena Steering BurdenAgents-1.316.6+8.9
Artificial Analysis Progress Index v1Indexes320345+8.8
MKA (Artificial Analysis)Professional Work35.343.8+8.5
TxBench Antibody Discovery V1 (100 evaluations)Science49.040.7−8.3
TxBench-Antibody-Development V1 (60 evaluations)Science15.022.8+7.8
Epoch Capabilities Index (ECI)Indexes159168+7.6
NoCoT-Bench NCKI (kspine v3)General QA121110−7.6
LMArena Agent Arena Praise vs ComplaintAgents4.920.0+7.6
VIALS-AA (pass@3)Multimodal64.071.4+7.5
Artificial Analysis Intelligence Index v4.3.2Indexes40.948.1+7.2
Vals Finance IndexIndexes60.553.3−7.1
Harvey LAB (Mercor run, undisclosed ~213-task cohort)Legal11.44.5−6.9
NeuroscienceBench V1 (91 evaluations)Science15.822.3+6.6
FrontierCode 1.1 Main (Artificial Analysis)Code37.143.7+6.6
BuseyBench SVGMultimodal6.67.3+6.5
APEX-Agents 1.1Professional Work72.979.3+6.4
DAXBench v2Code87.293.6+6.4
LLM Chess EloGames15461391−6.4
GDP.pdf (Artificial Analysis, All-pass)Professional Work22.828.6+5.8
Vibe Code Bench v1.1Code78.784.0+5.3
DualEntry Accounting AI BenchmarkFinance & Accounting83.278.2−5.0
APEX-Accounting 1.1Finance & Accounting58.863.7+4.9
DistressBenchSafety70.765.9−4.8
TxBench Oligo Discovery V1 (113 evaluations)Science35.130.4−4.7
APEX-1 (Mercor, mean score)Professional Work68.864.2−4.6
Vals Index v2.2Indexes56.060.5+4.6
BioSecBench-Surveillance V1 Clean RerunsScience44.139.9−4.2
LMArena Agent Arena Confirmed SuccessAgents1.39.7+4.2
Humanity's Last Exam (Text-Only)General QA42.146.2+4.1
Toloka Arena (v3)Agents59.055.0−4.0
BioSecBench-RespiratoryVirus-Surveillance V1 (110 evaluations)Science7.311.2+3.9
GDPval-AA v2.1Professional Work14421353−3.9
Blueprint-Bench 2Multimodal0.40.4+3.5
VariantBench V1 Clean RerunsScience31.434.8+3.4
AA-Briefcase v1.1 EloAgents12041265+3.3
BioSecBench-Function V1 Clean RerunsScience36.039.0+3.1
ChessBench AccuracyGames81.178.3−3.0
OncologyBench Strict (25 tasks, October 9 rubric)Health & Medicine13.316.0+2.7
NoCoT-Bench NCRI 15.2General QA126130+2.6
Visa Agentic Commerce BenchmarkAgents53.351.1−2.2
EpiBench V1 Clean RerunsScience21.423.3+1.9
LMArena Agent Arena Bash ReliabilityAgents9.913.4+1.8
SciCode 1.0.1 (Artificial Analysis)Code56.655.1−1.5
AA-LCR v1.1Search84.082.7−1.3
ScBench-Long V1 Clean RerunsScience19.918.8−1.1
OTIS Mock AIME 2024-2025Math98.9100.0+1.1
TxBench Preclinical Pharmacology V1 Clean RerunsScience54.755.7+1.0
AutomationBench-AA 1.0.6Agents60.960.2−0.7
Excel Modeling Benchmark v1.1Finance & Accounting72.973.3+0.5
SWE Atlas v1.1 - OverallCode57.357.7+0.4
Prophet Arena Arena Score (Agentic, OpenCode)Forecasting86.085.7−0.3
LMArena Agent Arena Tool HallucinationAgents0.41.0+0.3
MMMU ProMultimodal84.584.7+0.2
Design Arena Overall Code CategoriesCode12991300+0.2
AA-Omniscience IndexIndexes29.629.8+0.1
Prophet Arena Arena Score (Fixed Context)Forecasting85.385.20.0

Gemini 3.8 FlashvsGPT-6.1 Terranormalised score marginsData Nov 15, 2026Data: Artificial Analysis, benchmarks.bio, Vals AI and 24 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…