Atlas

Compare Models

Claude Fable 5.2 vs GPT-5.6 Terra

Claude Fable 5.2 leads by 9.5 capability points (≈95% interval +6.9 to +12.2): a clear lead over GPT-5.6 Terra · P(higher) ≈ 100%

Wins 50 of 51 shared benchmarks (0 lost, 1 tied)

Capability vs. relative cost
Claude Fable 5.2GPT-5.6 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-5.6 Terracapability vs. relative costData Nov 15, 2026Village LLM Atlasatlas.village.dev

Benchmark wins50 – 0 · 51 shared, 1 tied

Comparing matching setups only (6 benchmarks excluded) · suites counted once
← Claude Fable 5.2GPT-5.6 Terra →

Claude Fable 5.2vsGPT-5.6 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-5.6 Terra minus Claude Fable 5.2.

ARC-AGI-3 (Semi-Private)General QA81.20.7−80.5
LMArena Agent Arena Net Improvement (Steering Burden)Agents30.4-1.8−60.8
Mystery Game PuzzlesGames90.035.0−55.0
Humanity's Last Exam DiamondGeneral QA80.228.8−51.4
GameBench 2Games99.153.5−45.6
HardGeoBench Avg DistanceMultimodal5521991−44.4
ChartographyMultimodal73.434.0−39.4
LMArena Agent Arena Praise vs ComplaintAgents73.4-1.9−37.7
Humanity's Last ExamGeneral QA73.235.9−37.3
Furniture Assembly BenchmarkMultimodal90.054.2−35.8
Chess Puzzles (task version 2.0)Games79.047.0−32.0
AutomationBench 1.0.6 (Zapier)Agents44.314.8−29.5
FrontierMath Tier 4 (v2) - PrivateMath100.070.7−29.3
ProofBench v1.1Math100.074.0−26.0
HardGeoBench Country AccuracyMultimodal70.045.0−25.0
FrontierCode 2.0 (Artificial Analysis)Code48.823.9−24.9
AA-Briefcase v1.1 EloAgents17851333−24.8
HardGeoBench <=100miMultimodal60.036.3−23.7
HardGeoBench <=10miMultimodal48.825.0−23.7
HieroglyphBenchMultimodal43.620.5−23.1
Conceptual Reasoning IndexIndexes79.965.7−19.5
Vibe Code Bench v1.1Code93.874.6−19.2
τ³-Banking (Mercor run, bm25_grep retrieval)Agents55.738.8−16.9
LMArena Agent Arena Steering BurdenAgents33.0-0.5−16.7
Medical Long Context Reasoning (MLCR-AA, October 2026 methodology)Health & Medicine43.927.2−16.7
CritPt v2 (agentic, avg@5)Science86.369.7−16.6
GDPval-AA v2.1Professional Work18311461−16.3
APEX-Agents 1.1Professional Work87.671.8−15.8
LMCAGeneral QA70.655.0−15.7
ACCoRDGeneral QA89.474.7−14.7
LMArena Agent Arena Confirmed SuccessAgents22.0-7.2−14.6
AutomationBench-AA 1.0.6Agents74.259.6−14.5
DeepSWE v1.1Code74.360.2−14.1
SciCode 1.0.1 (Artificial Analysis)Code67.755.0−12.7
HealthBench Professional (Length-Adjusted)Health & Medicine68.156.3−11.8
BuseyBench SVGMultimodal7.26.1−10.9
Epoch Capabilities Index (ECI)Indexes174161−10.8
APEX-Accounting 1.1Finance & Accounting70.459.8−10.6
ARC-AGI-2General QA93.883.9−9.8
FrontierMath Tiers 1-3 (v2) - PrivateMath95.186.0−9.1
Harvey LAB (Mercor run, undisclosed ~213-task cohort)Legal10.02.3−7.7
LMArena Agent Arena Bash ReliabilityAgents14.40.7−6.8
GPQA DiamondGeneral QA93.987.4−6.5
DTBenchGeneral QA99.293.3−5.9
CritPt (avg@5)Science35.730.0−5.7
AA-LCR v1.1Search83.379.0−4.3
SWE-bench MultilingualCode96.694.4−2.2
ARC-AGI-1 Semi-PrivateGeneral QA98.596.5−2.0
Tax Agent BenchFinance & Accounting67.065.2−1.8
LMArena Agent Arena Tool HallucinationAgents1.01.00.0
OTIS Mock AIME 2024-2025Math99.799.70.0

Claude Fable 5.2vsGPT-5.6 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…