Atlas

Compare Models

Claude Fable 5.2 vs GPT-6.1 Terra

Claude Fable 5.2 leads by 4.9 capability points (≈95% interval +2.1 to +7.8): a clear lead over GPT-6.1 Terra · P(higher) ≈ 100%

Wins 44 of 51 shared benchmarks (6 lost, 1 tied)

Capability vs. relative cost
Claude Fable 5.2GPT-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 Fable 5.2vsGPT-6.1 Terracapability vs. relative costData Nov 15, 2026Village LLM Atlasatlas.village.dev

Benchmark wins44 – 6 · 51 shared, 1 tied

Comparing matching setups only (6 benchmarks excluded) · suites counted once
← Claude Fable 5.2GPT-6.1 Terra →

Claude Fable 5.2vsGPT-6.1 Terrawins by categoryData Nov 15, 2026Data: Artificial Analysis, Epoch AI, Arena and 13 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 Fable 5.2.

SimpleQA Verified (Epoch anti-abstention)General QA94.057.3−36.7
LMArena Agent Arena Net Improvement (Steering Burden)Agents30.412.0−34.7
Humanity's Last Exam DiamondGeneral QA80.247.3−32.9
Humanity's Last Exam (Text-Only)General QA74.346.2−28.0
LMArena Agent Arena Praise vs ComplaintAgents73.420.0−26.7
AA-Briefcase v1.1 EloAgents17851401−21.0
FrontierCode 1.1 Main (Artificial Analysis)Code64.043.7−20.3
τ³-Banking (Mercor run, bm25_grep retrieval)Agents55.737.1−18.6
HardGeoBench Avg DistanceMultimodal5521091−16.6
Mystery Game PuzzlesGames90.074.0−16.0
GDPval-AA v2.1Professional Work18311476−15.7
BioSecBench-Structure V1 (58 evaluations)Science41.425.9−15.5
AutomationBench-AA 1.0.6Agents74.260.2−14.0
HardGeoBench <=100miMultimodal60.046.3−13.7
HardGeoBench <=10miMultimodal48.835.0−13.7
Finance Agent v2.1Finance & Accounting69.556.0−13.5
Medical Long Context Reasoning (MLCR-AA, October 2026 methodology)Health & Medicine43.930.6−13.3
GameBench 2Games99.185.8−13.3
SciCode 1.0.1 (Artificial Analysis)Code67.755.1−12.6
EBR-bench v4 (card ban, single agent)Games90.578.1−12.4
Tax Agent BenchFinance & Accounting67.055.0−12.0
HardGeoBench Country AccuracyMultimodal70.058.8−11.2
Furniture Assembly BenchmarkMultimodal90.079.2−10.8
Vals Finance IndexIndexes63.653.3−10.3
ChartographyMultimodal73.463.2−10.2
Vibe Code Bench v1.1Code93.884.0−9.8
APEX-Agents 1.1Professional Work87.679.3−8.3
LMArena Agent Arena Steering BurdenAgents33.016.6−8.2
GDPxProfessional Work16.48.5−8.0
CritPt (avg@5)Science35.728.0−7.7
FrontierMath Tier 4 (v2) - PrivateMath100.092.7−7.3
APEX-Accounting 1.1Finance & Accounting70.463.7−6.7
FrontierCode 2.0 (Artificial Analysis)Code48.842.3−6.4
MacroscopeBenchCode76.082.3+6.3
LMArena Agent Arena Confirmed SuccessAgents22.09.7−6.2
Harvey LAB (Mercor run, undisclosed ~213-task cohort)Legal10.04.5−5.5
ChessBenchGames69.364.0−5.3
DAXBench v2Code88.393.6+5.3
Epoch Capabilities Index (ECI)Indexes174168−5.0
CritPt v2 (agentic, avg@5)Science86.381.4−4.9
FrontierMath Tiers 1-3 (v2) - PrivateMath95.190.9−4.2
Excel Modeling Benchmark v1.1Finance & Accounting77.473.3−4.0
HLAB-Vals-1.1Legal5.01.7−3.3
DualEntry Accounting AI BenchmarkFinance & Accounting81.278.2−3.0
SpireClimbGames19.622.3+2.7
BuseyBench SVGMultimodal7.27.3+1.1
AA-LCR v1.1Search83.382.7−0.7
LMArena Agent Arena Bash ReliabilityAgents14.413.4−0.5
GPQA DiamondGeneral QA93.994.3+0.4
OTIS Mock AIME 2024-2025Math99.7100.0+0.3
LMArena Agent Arena Tool HallucinationAgents1.01.00.0

Claude Fable 5.2vsGPT-6.1 Terranormalised score marginsData Nov 15, 2026Data: Artificial Analysis, Epoch AI, Arena and 13 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…