Atlas

Compare Models

Claude Haiku 5.5 vs GPT-6.1 Terra

GPT-6.1 Terra leads by 4.0 capability points (≈95% interval +1.7 to +6.3): a clear lead over Claude Haiku 5.5 · P(higher) ≈ 100%

Wins 72 of 92 shared benchmarks (19 lost, 1 tied)

Capability vs. relative cost
Claude Haiku 5.5GPT-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.

Claude Haiku 5.5vsGPT-6.1 Terracapability vs. relative costData Nov 15, 2026Village LLM Atlasatlas.village.dev

Benchmark wins19 – 72 · 92 shared, 1 tied

Comparing matching setups only (18 benchmarks excluded) · suites counted once
← Claude Haiku 5.5GPT-6.1 Terra →

Claude Haiku 5.5vsGPT-6.1 Terrawins by categoryData Nov 15, 2026Data: Artificial Analysis, Vals AI, Epoch AI and 22 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 Claude Haiku 5.5.

FrontierMath Tier 4 (v2) - PrivateMath46.392.7+46.3
Mystery Game PuzzlesGames30.074.0+44.0
VoxelBench Text RatingGames16962313+43.9
Terminal-Bench-Science 0.1Science20.057.6+37.6
EBR-bench v4 (card ban, single agent)Games43.878.1+34.3
NoCoT-Bench NCRI 15.2General QA68.3130+34.1
Toloka Arena (v3)Agents21.055.0+34.0
SimpleQA Verified (Epoch anti-abstention)General QA23.857.3+33.5
Vending-Bench 2Games36029107+33.3
Furniture Assembly BenchmarkMultimodal47.579.2+31.7
ALE-BenchCode19962875+31.1
BuseyBench SVGMultimodal4.67.3+27.2
AutomationBench-AA 1.0.6Agents36.060.2+24.2
Terminal-Bench 4.0 (Artificial Analysis)Code32.855.6+22.7
ChessBenchGames41.364.0+22.7
ChessBench EloGames8751251+22.4
TaxCalcBench TY25Finance & Accounting2.024.0+22.0
Code Migration v1.1 (Vals Index v2.2 subset)Code50.272.2+22.0
GBENCH One-shot Coding GScoreGames55.677.6+21.9
LMArena Agent Arena Net Improvement (Steering Burden)Agents0.512.0+21.8
Legal Research BenchLegal43.826.0−17.8
Humanity's Last Exam DiamondGeneral QA29.647.3+17.7
CritPt v2 (agentic, avg@5)Science64.081.4+17.4
NoCoT-Bench NCKI (kspine v3)General QA85.9110+16.5
FrontierMath Tiers 1-3 (v2) - PrivateMath75.190.9+15.8
Artificial Analysis Intelligence Index v5.0.1Indexes36.551.3+14.8
Artificial Analysis Intelligence Index v5.0Indexes36.251.0+14.7
Terminal-Bench 4.0Code36.449.5+13.1
ChessBench AccuracyGames66.978.3+12.2
Artificial Analysis Engineering IndexIndexes42.554.7+12.2
LMArena Agent Arena Praise vs ComplaintAgents-3.520.0+11.8
DAXBench v2Code81.993.6+11.7
Vals Legal IndexIndexes70.358.8−11.5
VIALS-AA (pass^3)Multimodal49.160.2+11.2
APEX-Agents 1.1Professional Work68.379.3+11.0
Artificial Analysis Strategy & Ops IndexIndexes39.150.0+10.9
GameBench 2Games75.185.8+10.8
LLM Chess EloGames11431391+10.2
AA-Omniscience IndexIndexes10.729.8+9.6
Finance Agent v2.1Finance & Accounting65.456.0−9.4
AA-Briefcase v1.1 EloAgents15731401−9.4
Blueprint-Bench 2Multimodal0.30.4+9.1
FORTRESS Average Risk Score (ARS)Safety9.518.2−8.7
Excel Modeling Benchmark v1.1Finance & Accounting65.573.3+7.9
GDP.pdf (Artificial Analysis, All-pass)Professional Work20.828.6+7.8
Artificial Analysis Legal IndexIndexes38.746.4+7.7
HardGeoBench Country AccuracyMultimodal51.358.8+7.5
FrontierCode 2.0 (Artificial Analysis)Code34.942.3+7.4
Artificial Analysis Healthcare & Medical IndexIndexes37.544.7+7.2
Tax Agent BenchFinance & Accounting62.255.0−7.2
LMArena Agent Arena Steering BurdenAgents2.416.6+7.1
HardGeoBench Avg DistanceMultimodal13191091+7.0
GDPval-AA v2.1Professional Work16291476−6.7
Artificial Analysis Progress Index v1Indexes326345+6.7
MultiChallengeChat & Writing67.774.2+6.5
Vibe Code Bench v1.1Code90.484.0−6.4
Epoch Capabilities Index (ECI)Indexes160168+6.4
τ³-Banking (Mercor run, bm25_grep retrieval)Agents43.037.1−5.9
Artificial Analysis Economics IndexIndexes42.748.5+5.8
DistressBenchSafety71.765.9−5.8
VIALS-AA (average of 3)Multimodal60.566.0+5.6
MMMU ProMultimodal79.184.7+5.6
Artificial Analysis Finance & Accounting IndexIndexes43.649.0+5.4
CritPt (avg@5)Science22.628.0+5.4
HardGeoBench <=10miMultimodal30.035.0+5.0
Artificial Analysis Intelligence Index v4.3.2Indexes43.448.1+4.7
Medical Long Context Reasoning (MLCR-AA, October 2026 methodology)Health & Medicine35.030.6−4.4
AA-LCR v1.1Search78.382.7+4.3
Vals Finance IndexIndexes57.453.3−4.1
FrontierCode 1.1 Main (Artificial Analysis)Code45.541.4−4.1
ChessBench CoherenceGames80.784.8+4.1
LMArena Agent Arena Confirmed SuccessAgents1.69.7+4.0
LiveBench 2.0Agents75.679.7+4.0
Humanity's Last Exam (Text-Only)General QA42.746.2+3.6
GDPxProfessional Work4.98.5+3.5
Design Arena Overall Code CategoriesCode12801300+3.3
LMArena Agent Arena Bash ReliabilityAgents7.413.4+3.0
MKA (Artificial Analysis)Professional Work41.043.8+2.8
HLAB-Vals-1.1Legal4.21.7−2.5
DualEntry Accounting AI BenchmarkFinance & Accounting76.278.2+2.0
Vals Index v2.2Indexes58.760.5+1.8
SWE Atlas v1.1 - OverallCode59.457.7−1.7
LMArena Agent Arena Tool HallucinationAgents-2.11.0+1.5
APEX-Accounting 1.1Finance & Accounting62.263.7+1.5
HardGeoBench <=100miMultimodal45.046.3+1.3
OTIS Mock AIME 2024-2025Math98.9100.0+1.1
APEX-1 (Mercor, mean score)Professional Work64.764.2−0.5
Harvey LAB (Mercor run, undisclosed ~213-task cohort)Legal4.94.5−0.4
Prophet Arena Arena Score (Agentic, OpenCode)Forecasting85.985.7−0.2
Prophet Arena Arena Score (Fixed Context)Forecasting85.485.2−0.2
SciCode 1.0.1 (Artificial Analysis)Code55.055.1+0.1
VIALS-AA (pass@3)Multimodal71.471.40.0

Claude Haiku 5.5vsGPT-6.1 Terranormalised score marginsData Nov 15, 2026Data: Artificial Analysis, Vals AI, Epoch AI and 22 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…