Atlas

Compare Models

Claude Fable 5.1 vs Gemini 4 Carbon

Gemini 4 Carbon leads by 2.2 capability points (≈95% interval −0.5 to +4.9): statistically tied · P(higher) ≈ 94%

Wins 34 of 40 shared benchmarks (6 lost)

Capability vs. relative cost
Claude Fable 5.1Gemini 4 CarbonClosed 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 5.1vsGemini 4 Carboncapability vs. relative costData Nov 15, 2026Village LLM Atlasatlas.village.dev

Benchmark wins6 – 34 · 40 shared

Comparing matching setups only (8 benchmarks excluded) · suites counted once
← Claude Fable 5.1Gemini 4 Carbon →

Claude Fable 5.1vsGemini 4 Carbonwins by categoryData Nov 15, 2026Data: Arena, Mercor, Artificial Analysis and 15 othersVillage LLM Atlasatlas.village.dev

Margins

Each benchmark scaled 0–100% (its declared range, else the field’s range); Δ is Gemini 4 Carbon minus Claude Fable 5.1.

TxBench-Antibody-Development V1 (60 evaluations)Science27.861.1+33.3
AutomationBench 1.0.6 (Zapier)Agents22.453.0+30.6
ChartographyMultimodal46.272.5+26.3
APEX-Agents 1.1Professional Work71.890.4+18.6
FoodTruck BenchGames6273277218+18.6
Blueprint-Bench 2Multimodal0.40.6+16.7
RiemannBenchMath65.680.0+14.4
Bug Hunt BenchCode41.054.6+13.7
CWE-bench v1 Programmatic Pass@1Cybersecurity58.071.0+13.0
FrontierCode 2.0 - Pass RateCode37.646.9+9.3
CursorBench 4.0Code51.860.4+8.6
FrontierCode 2.0 - ScoreCode34.042.6+8.6
Vals RSI Index v1.1 (four tasks)Agents36.127.7−8.4
BrokenArXiv 08/2026 (Native Harnesses)Math79.888.1+8.3
BuseyBench SVGMultimodal7.07.9+8.3
AA-LCR v1.1Search87.579.3−8.2
LMArena Agent Arena Bash ReliabilityAgents10.1-5.2−7.6
Harvey LAB (Mercor run, undisclosed ~213-task cohort)Legal17.423.0+5.6
ArXivMath 08/2026 (Native Harnesses)Math87.793.0+5.3
LMArena Agent Arena Praise vs ComplaintAgents31.641.0+4.7
CritPt v2 (agentic, avg@5)Science77.782.3+4.6
ARC-AGI-2General QA90.094.2+4.2
LMArena Agent Arena Net Improvement (Steering Burden)Agents14.817.0+4.1
LMArena Text Arena EloChat & Writing15031526+4.1
APEX-Accounting 1.1Finance & Accounting66.770.7+4.0
GDPval-AA v2.1Professional Work16441733+3.9
OpenWorldEngineer (Mean@8)Code29.132.8+3.6
LMArena Code Arena WebDevCode17421768+3.4
MMMU ProMultimodal86.189.2+3.1
GDPxProfessional Work8.510.9+2.3
LMArena Agent Arena Steering BurdenAgents14.819.1+2.1
DAXBench v2Code91.593.6+2.1
LMArena Agent Arena Confirmed SuccessAgents14.418.4+2.0
AA-Briefcase v1.1 EloAgents15771611+1.9
τ³-Banking (Mercor run, bm25_grep retrieval)Agents54.055.7+1.7
Tax Agent BenchFinance & Accounting77.976.4−1.5
Medical Long Context Reasoning (MLCR-AA, October 2026 methodology)Health & Medicine18.319.4+1.1
ARC-AGI-1 Semi-PrivateGeneral QA97.598.5+1.0
Excel Modeling Benchmark v1.1Finance & Accounting76.776.5−0.3
LMArena Agent Arena Tool HallucinationAgents0.80.5−0.1

Claude Fable 5.1vsGemini 4 Carbonnormalised score marginsData Nov 15, 2026Data: Arena, Mercor, Artificial Analysis and 15 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…