Atlas

Compare Models

Claude Fable 5 vs GPT-5.6 Sol Pro

GPT-5.6 Sol Pro leads by 0.1 capability points (≈95% interval −2.5 to +2.7): statistically tied · P(higher) ≈ 53%

Claude Fable 5 wins 14 of 23 shared benchmarks (8 lost, 1 tied)

Capability vs. relative cost
Claude Fable 5GPT-5.6 Sol ProClosed 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 Sol Pro has no setting with a relative cost yet.

Claude Fable 5vsGPT-5.6 Sol Procapability vs. relative costData Nov 15, 2026Village LLM Atlasatlas.village.dev

Benchmark wins14 – 8 · 23 shared, 1 tied

Comparing matching setups only (93 benchmarks excluded) · suites counted once
← Claude Fable 5GPT-5.6 Sol Pro →

Claude Fable 5vsGPT-5.6 Sol Prowins by categoryData Nov 15, 2026Data: Mercor, MineBench Maintainers, OpenAI and 7 othersVillage LLM Atlasatlas.village.dev

Margins

Each benchmark scaled 0–100% (its declared range, else the field’s range); Δ is GPT-5.6 Sol Pro minus Claude Fable 5.

SWE Atlas Codebase QnA Extended (Mercor held-out)Code0.024.7+24.7
SWE-bench VerifiedCode95.572.8−22.7
GeneBench-Pro Public Release SubsetScience14.034.0+20.0
SWE-bench Verified Extended (Mercor held-out)Code78.763.3−15.4
BrowseComp (Mercor 130-question subset)Search82.595.7+13.2
Terminal-Bench 4.0Code44.532.3−12.2
Harvey LAB (Mercor run, 1,749 public tasks)Legal14.42.7−11.7
FrontierMath Tier 4 (v2) - PrivateMath90.280.5−9.7
MineBench — Bradley-Terry RatingGames18691994+7.6
APEX-1 (Mercor, mean score)Professional Work66.073.1+7.1
Terminal-Bench 2.1Code88.081.2−6.8
MineBench Conservative RatingGames18872000+6.7
APEX-AccountingFinance & Accounting56.451.5−4.9
HealthBench Professional (Length-Adjusted)Health & Medicine66.061.4−4.6
RedlineBenchLegal42.445.8+3.4
APEX-AgentsProfessional Work43.340.0−3.3
WordleBenchGames97.094.0−3.0
WeirdML (v2)Science91.989.4−2.5
SWE Atlas - Codebase QnACode33.130.9−2.2
Ruby LLM Benchmarks — Program FixerCode77.275.3−1.9
GPQA DiamondGeneral QA94.194.9+0.8
SWE-bench MultilingualCode94.894.2−0.6
Eval ConnectionsGames100.0100.00.0

Claude Fable 5vsGPT-5.6 Sol Pronormalised score marginsData Nov 15, 2026Data: Mercor, MineBench Maintainers, OpenAI and 7 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…