Atlas

Compare Models

Claude Fable 5.2 vs Gemini 3.7 Flash

Claude Fable 5.2 leads by 10.9 capability points (≈95% interval +8.3 to +13.4): a clear lead over Gemini 3.7 Flash · P(higher) ≈ 100%

Wins 47 of 48 shared benchmarks (1 lost)

Capability vs. relative cost
Claude Fable 5.2Gemini 3.7 FlashClosed 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 5.2vsGemini 3.7 Flashcapability vs. relative costData Nov 15, 2026Village LLM Atlasatlas.village.dev

Benchmark wins47 – 1 · 48 shared

Comparing matching setups only (5 benchmarks excluded) · suites counted once
← Claude Fable 5.2Gemini 3.7 Flash →

Claude Fable 5.2vsGemini 3.7 Flashwins by categoryData Nov 15, 2026Data: Artificial Analysis, Epoch AI, Arena and 14 othersVillage LLM Atlasatlas.village.dev

Margins

Each benchmark scaled 0–100% (its declared range, else the field’s range); Δ is Gemini 3.7 Flash minus Claude Fable 5.2.

FrontierMath Tier 4 (v2) - PrivateMath100.036.6−63.4
Furniture Assembly BenchmarkMultimodal90.026.7−63.3
LMArena Agent Arena Net Improvement (Steering Burden)Agents30.4-2.2−61.5
Mystery Game PuzzlesGames90.037.0−53.0
FrontierSWE V2Code69.920.3−49.6
Humanity's Last Exam DiamondGeneral QA80.231.2−49.0
Arena Code WebDev — ReactCode19071584−43.3
ProofBench v1.1Math100.058.0−42.0
LMArena Agent Arena Praise vs ComplaintAgents73.4-2.9−38.2
Remote Labor Index (RLI)Professional Work41.35.0−36.3
Bug Hunt BenchCode55.221.0−34.3
ChartographyMultimodal73.440.4−33.0
Medical Long Context Reasoning (MLCR-AA, October 2026 methodology)Health & Medicine43.912.2−31.7
AA-Briefcase v1.1 EloAgents16771115−30.8
Humanity's Last ExamGeneral QA73.245.6−27.6
SimpleQA Verified (Epoch anti-abstention)General QA94.069.2−24.8
FrontierMath Tiers 1-3 (v2) - PrivateMath95.171.6−23.5
Vibe Code Bench v1.1Code93.870.4−23.4
Arena Code WebDev — HTMLCode17501593−22.0
CritPt (avg@5)Science35.714.3−21.4
Conceptual Reasoning IndexIndexes79.964.6−21.1
LMCAGeneral QA70.650.4−20.2
LMArena Agent Arena Steering BurdenAgents33.0-7.3−20.2
HealthBench Professional (Length-Adjusted)Health & Medicine68.148.4−19.7
Vals RSI Index v1.1 (four tasks)Agents37.918.3−19.6
SWE-bench MultilingualCode96.682.6−14.0
GDPval-AA v2.1Professional Work17181404−13.8
APEX-Accounting 1.1Finance & Accounting70.457.2−13.2
Chess Puzzles (task version 2.0)Games79.066.0−13.0
Epoch Capabilities Index (ECI)Indexes174158−12.8
LMArena Agent Arena Confirmed SuccessAgents22.0-3.5−12.7
ACCoRDGeneral QA89.477.0−12.4
AutomationBench-AA 1.0.6Agents74.262.0−12.1
HieroglyphBenchMultimodal43.655.0+11.4
MineBench — Bradley-Terry RatingGames19321765−10.1
Ramp Accounting BenchFinance & Accounting52.242.3−9.9
Tax Agent BenchFinance & Accounting67.057.7−9.3
ARC-AGI-2General QA93.884.6−9.2
DeepSWE v1.1Code74.365.5−8.8
APEX-Agents 1.1Professional Work87.679.4−8.2
SciCode 1.0.1 (Artificial Analysis)Code67.759.8−7.9
BuseyBench SVGMultimodal7.26.6−5.3
LMArena Agent Arena Bash ReliabilityAgents14.46.2−4.1
ARC-AGI-1 Semi-PrivateGeneral QA98.595.5−3.0
OTIS Mock AIME 2024-2025Math99.797.2−2.5
DTBenchGeneral QA99.296.8−2.4
AA-LCR v1.1Search83.383.0−0.3
LMArena Agent Arena Tool HallucinationAgents1.00.7−0.2

Claude Fable 5.2vsGemini 3.7 Flashnormalised score marginsData Nov 15, 2026Data: Artificial Analysis, Epoch AI, Arena and 14 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…