Atlas

Compare Models

GPT-5.6 Terra vs GPT-6.1 Terra

GPT-6.1 Terra leads by 4.6 capability points (≈95% interval +2.3 to +6.9): a clear lead over GPT-5.6 Terra · P(higher) ≈ 100%

Wins 64 of 75 shared benchmarks (11 lost)

Capability vs. relative cost
GPT-5.6 TerraGPT-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.

GPT-5.6 TerravsGPT-6.1 Terracapability vs. relative costData Nov 15, 2026Village LLM Atlasatlas.village.dev

Benchmark wins11 – 64 · 75 shared

Comparing matching setups only (2 benchmarks excluded) · suites counted once
← GPT-5.6 TerraGPT-6.1 Terra →

GPT-5.6 TerravsGPT-6.1 Terrawins by categoryData Nov 15, 2026Data: Artificial Analysis, Arena, OpenAI and 21 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 GPT-5.6 Terra.

Terminal-Bench-Science 0.1Science13.857.6+43.8
Mystery Game PuzzlesGames35.074.0+39.0
GameBench 2Games53.585.8+32.4
Arena Code WebDev — HTMLCode15231738+30.1
ChartographyMultimodal34.063.2+29.2
HardGeoBench Avg DistanceMultimodal19911091+27.8
Terminal-Bench 4.0Code22.749.5+26.8
LMArena Agent Arena Net Improvement (Steering Burden)Agents-1.812.0+26.1
ALE-BenchCode21542875+25.5
Furniture Assembly BenchmarkMultimodal54.279.2+25.0
NoCoT-Bench NCRI 15.2General QA89.0130+22.8
FrontierMath Tier 4 (v2) - PrivateMath70.792.7+22.0
Arena Code WebDev — ReactCode15111665+20.7
Terminal-Bench 4.0 (Artificial Analysis)Code35.455.6+20.2
ChessBench EloGames9251251+19.4
Artificial Analysis Intelligence Index v5.0.1Indexes32.351.3+19.0
Humanity's Last Exam DiamondGeneral QA28.847.3+18.5
FrontierCode 2.0 (Artificial Analysis)Code23.942.3+18.4
Legal Research BenchLegal41.326.0−15.4
ChessBench CoherenceGames69.784.8+15.1
AA-Omniscience IndexIndexes0.129.8+14.9
GBENCH One-shot Coding GScoreGames63.377.6+14.3
HardGeoBench Country AccuracyMultimodal45.058.8+13.8
BuseyBench SVGMultimodal6.17.3+12.0
CritPt v2 (agentic, avg@5)Science69.781.4+11.7
Agents' Last Exam Overall ScoreAgents48.560.0+11.5
Blueprint-Bench 2Multimodal0.30.4+11.3
LMArena Agent Arena Praise vs ComplaintAgents-1.920.0+11.0
FrontierPhysics v0.1Science3.614.3+10.7
Vending-Bench 2Games73439107+10.7
DistressBenchSafety76.465.9−10.5
Design Arena Overall Code CategoriesCode12381300+10.3
Tax Agent BenchFinance & Accounting65.255.0−10.2
HardGeoBench <=100miMultimodal36.346.3+10.0
HardGeoBench <=10miMultimodal25.035.0+10.0
Vibe Code Bench v1.1Code74.684.0+9.4
MentalHealthBenchSafety46.255.5+9.3
LMArena Agent Arena Steering BurdenAgents-0.516.6+8.6
LMArena Agent Arena Confirmed SuccessAgents-7.29.7+8.4
HealthBench Hard (updated implementation)Health & Medicine34.326.0−8.3
Bug Hunt BenchCode30.538.7+8.3
HealthBench (updated implementation)Health & Medicine58.751.1−7.6
Artificial Analysis Progress Index v1Indexes324345+7.5
APEX-Agents 1.1Professional Work71.879.3+7.5
GPQA DiamondGeneral QA87.494.3+6.9
ChessBench AccuracyGames71.978.3+6.9
LMArena Agent Arena Bash ReliabilityAgents0.713.4+6.3
Artificial Analysis Intelligence Index v4.3.2Indexes42.148.1+6.0
Epoch Capabilities Index (ECI)Indexes161168+5.9
BioSecBench-RespiratoryVirus-Surveillance V1 (110 evaluations)Science5.811.2+5.5
APEX-1 (Mercor, mean score)Professional Work69.564.2−5.3
MMMU ProMultimodal79.584.7+5.2
FrontierMath Tiers 1-3 (v2) - PrivateMath86.090.9+4.9
LiveBench 2.0Agents75.079.7+4.7
Humanity's Last Exam (Text-Only)General QA41.946.2+4.4
NoCoT-Bench NCKI (kspine v3)General QA104110+4.1
LLM Chess EloGames12931391+4.0
GDP.pdf (Artificial Analysis, All-pass)Professional Work24.628.6+4.0
APEX-Accounting 1.1Finance & Accounting59.863.7+3.9
AA-Briefcase v1.1 EloAgents13331401+3.7
AA-LCR v1.1Search79.082.7+3.7
Medical Long Context Reasoning (MLCR-AA, October 2026 methodology)Health & Medicine27.230.6+3.3
Harvey LAB (Mercor run, undisclosed ~213-task cohort)Legal2.34.5+2.2
CritPt (avg@5)Science30.028.0−2.0
HealthBench (length-adjusted)Health & Medicine57.055.1−1.9
HealthBench ProfessionalHealth & Medicine62.464.2+1.8
τ³-Banking (Mercor run, bm25_grep retrieval)Agents38.837.1−1.7
HealthBench Hard (length-adjusted)Health & Medicine32.731.4−1.3
VulcanBench Frontier v4Code89.190.0+0.9
GDPval-AA v2.1Professional Work14611476+0.6
AutomationBench-AA 1.0.6Agents59.660.2+0.5
HealthBench Consensus (updated implementation)Health & Medicine95.294.8−0.4
OTIS Mock AIME 2024-2025Math99.7100.0+0.3
SciCode 1.0.1 (Artificial Analysis)Code55.055.1+0.1
LMArena Agent Arena Tool HallucinationAgents1.01.00.0

GPT-5.6 TerravsGPT-6.1 Terranormalised score marginsData Nov 15, 2026Data: Artificial Analysis, Arena, OpenAI and 21 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…