Atlas

Compare Models

GPT-6.1 Terra vs Grok 4.7

GPT-6.1 Terra leads by 4.7 capability points (≈95% interval +2.4 to +7.0): a clear lead over Grok 4.7 · P(higher) ≈ 100%

Wins 64 of 89 shared benchmarks (25 lost)

Capability vs. relative cost
GPT-6.1 TerraGrok 4.7Closed 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.

GPT-6.1 TerravsGrok 4.7capability vs. relative costData Nov 15, 2026Village LLM Atlasatlas.village.dev

Benchmark wins64 – 25 · 89 shared

Comparing matching setups only (28 benchmarks excluded) · suites counted once
← GPT-6.1 TerraGrok 4.7 →

GPT-6.1 TerravsGrok 4.7wins by categoryData Nov 15, 2026Data: Artificial Analysis, Vals AI, Epoch AI and 26 othersVillage LLM Atlasatlas.village.dev

Margins

Each benchmark scaled 0–100% (its declared range, else the field’s range); Δ is Grok 4.7 minus GPT-6.1 Terra.

FrontierMath Tier 4 (v2) - PrivateMath92.717.1−75.6
Furniture Assembly BenchmarkMultimodal79.220.8−58.3
ChartographyMultimodal63.214.7−48.5
Mystery Game PuzzlesGames74.029.0−45.0
FrontierMath Tiers 1-3 (v2) - PrivateMath90.953.0−37.9
ALE-BenchCode28751823−37.3
HardGeoBench Avg DistanceMultimodal10912104−31.2
Terminal-Bench-Science 0.1Science41.911.4−30.5
BuseyBench SVGMultimodal7.34.6−27.1
Toloka Arena (v3)Agents55.031.0−24.0
Humanity's Last Exam DiamondGeneral QA47.323.6−23.7
LLM Chess EloGames1391822−23.4
VIALS-AA (pass^3)Multimodal60.237.9−22.4
GBENCH One-shot Coding GScoreGames77.656.3−21.3
Legal Research BenchLegal26.047.1+21.2
CritPt v2 (agentic, avg@5)Science76.955.7−21.1
Terminal-Bench 4.0 (Artificial Analysis)Code46.525.8−20.7
Code Migration v1.1 (Vals Index v2.2 subset)Code72.251.8−20.4
WeirdML v3Science28.18.5−19.6
HardGeoBench <=100miMultimodal46.327.5−18.7
LMArena Agent Arena Net Improvement (Steering Burden)Agents12.02.6−17.8
HardGeoBench <=10miMultimodal35.017.5−17.5
HLAB-Vals-1.1Legal1.718.3+16.7
AA-Briefcase v1.1 EloAgents13341636+16.5
TaxCalcBench TY25Finance & Accounting24.08.0−16.0
Artificial Analysis Intelligence Index v5.0.1Indexes51.335.5−15.8
Artificial Analysis Intelligence Index v5.0Indexes51.035.3−15.6
HardGeoBench Country AccuracyMultimodal58.843.8−15.0
Harvey LAB (Mercor run, undisclosed ~213-task cohort)Legal4.519.5+15.0
Tax Agent BenchFinance & Accounting55.069.0+14.0
Vals Legal IndexIndexes58.872.7+13.9
VIALS-AA (average of 3)Multimodal66.052.4−13.7
GameBench 2Games85.872.3−13.5
DistressBenchSafety65.952.4−13.5
GDPval-AA v2.1Professional Work14161717+13.2
Artificial Analysis Engineering IndexIndexes54.741.8−13.0
Terminal-Bench 4.0Code50.337.6−12.7
APEX-Agents 1.1Professional Work79.368.7−10.6
ChessBench EloGames12511076−10.4
CritPt (avg@5)Science28.018.0−10.0
DualEntry Accounting AI BenchmarkFinance & Accounting78.288.1+9.9
Epoch Capabilities Index (ECI)Indexes168156−9.7
Blueprint-Bench 2Multimodal0.40.3−9.6
Medical Long Context Reasoning (MLCR-AA, October 2026 methodology)Health & Medicine30.621.1−9.4
LMArena Agent Arena Steering BurdenAgents16.6-2.3−9.4
FrontierMemory v1.1Agents46.737.8−8.9
Visa Agentic Commerce BenchmarkAgents51.160.0+8.9
Vending-Bench 2Games910710537+8.7
LMArena Agent Arena Praise vs ComplaintAgents20.03.1−8.5
APEX-Accounting 1.1Finance & Accounting63.755.3−8.4
FrontierCode 2.0 (Artificial Analysis)Code36.928.7−8.2
Artificial Analysis Progress Index v1Indexes345323−7.9
MMMU ProMultimodal84.778.2−6.5
VIALS-AA (pass@3)Multimodal71.465.8−5.6
LiveBench 2.0Agents79.774.1−5.6
Vals Finance IndexIndexes53.358.8+5.5
GDP.pdf (Artificial Analysis, All-pass)Professional Work28.623.2−5.4
AutomationBench-AA 1.0.6Agents60.265.6+5.4
Design Arena Overall Code CategoriesCode13001270−5.0
Artificial Analysis Strategy & Ops IndexIndexes50.054.8+4.8
Finance Agent v2.1Finance & Accounting56.060.6+4.6
GDPxProfessional Work8.53.9−4.6
ChessBench AccuracyGames78.374.0−4.6
τ³-Banking (Mercor run, bm25_grep retrieval)Agents37.132.6−4.5
SWE Atlas v1.1 - OverallCode57.762.1+4.4
MKA (Artificial Analysis)Professional Work43.840.1−3.6
FrontierCode 1.1 Main (Artificial Analysis)Code36.339.6+3.3
Humanity's Last Exam (Text-Only)General QA46.243.1−3.1
Artificial Analysis Economics IndexIndexes48.545.7−2.8
Artificial Analysis Finance & Accounting IndexIndexes49.051.7+2.7
OncologyBench Strict (25 tasks, October 9 rubric)Health & Medicine16.018.7+2.7
AA-LCR v1.1Search82.780.0−2.7
SciCode 1.0.1 (Artificial Analysis)Code55.157.8+2.7
LMArena Agent Arena Confirmed SuccessAgents9.75.1−2.3
Vibe Code Bench v1.1Code84.086.2+2.2
LMArena Agent Arena Bash ReliabilityAgents13.49.3−2.0
OTIS Mock AIME 2024-2025Math100.098.1−1.9
Vals Index v2.2Indexes60.558.6−1.9
Artificial Analysis Healthcare & Medical IndexIndexes44.746.6+1.9
Excel Modeling Benchmark v1.1Finance & Accounting73.371.5−1.8
Artificial Analysis Intelligence Index v4.3.2Indexes48.146.4−1.7
GPQA DiamondGeneral QA94.392.7−1.6
ChessBench CoherenceGames84.886.3+1.5
SimpleQA Verified (Epoch anti-abstention)General QA57.356.0−1.3
APEX-1 (Mercor, mean score)Professional Work64.265.4+1.2
AA-Omniscience IndexIndexes29.832.0+1.1
DAXBench v2Code93.692.6−1.1
Artificial Analysis Legal IndexIndexes46.447.2+0.8
LMArena Agent Arena Tool HallucinationAgents1.01.00.0

GPT-6.1 TerravsGrok 4.7normalised score marginsData Nov 15, 2026Data: Artificial Analysis, Vals AI, Epoch AI and 26 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…