Atlas

Compare Models

Claude Fable 5.2 vs DeepSeek V4.1 Flash

Claude Fable 5.2 leads by 10.8 capability points (≈95% interval +8.3 to +13.4): a clear lead over DeepSeek V4.1 Flash · P(higher) ≈ 100%

Wins 57 of 65 shared benchmarks (8 lost)

Capability vs. relative cost
Claude Fable 5.2DeepSeek V4.1 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 5.2vsDeepSeek V4.1 Flashcapability vs. relative costData Nov 15, 2026Village LLM Atlasatlas.village.dev

Benchmark wins57 – 8 · 65 shared

Comparing matching setups only (5 benchmarks excluded) · suites counted once
← Claude Fable 5.2DeepSeek V4.1 Flash →

Claude Fable 5.2vsDeepSeek V4.1 Flashwins by categoryData Nov 15, 2026Data: Artificial Analysis, Vals AI, Epoch AI and 17 othersVillage LLM Atlasatlas.village.dev

Margins

Each benchmark scaled 0–100% (its declared range, else the field’s range); Δ is DeepSeek V4.1 Flash minus Claude Fable 5.2.

ARC-AGI-3 (Semi-Private)General QA92.33.3−89.0
FrontierMath Tier 4 (v2) - PrivateMath100.026.8−73.2
ChartographyMultimodal73.411.6−61.8
Humanity's Last Exam DiamondGeneral QA80.219.1−61.1
Furniture Assembly BenchmarkMultimodal90.034.2−55.8
ArXivMath 09/2026 (Native Harnesses)Math96.240.6−55.7
SimpleQA Verified (Epoch anti-abstention)General QA94.042.9−51.1
Mystery Game PuzzlesGames90.043.0−47.0
AA-PersonalityBench Organization indexChat & Writing54.3100+46.2
ProofBench v1.1Math100.054.0−46.0
AA-PersonalityBench Warmth indexChat & Writing82.941.0−42.3
OpenWorldEngineer (Mean@8)Code54.612.4−42.2
APEX-Agents 1.1Professional Work87.650.5−37.1
AA-PersonalityBench Imagination indexChat & Writing67.631.5−36.5
Humanity's Last ExamGeneral QA73.236.8−36.4
Bug Hunt BenchCode55.220.6−34.6
AA-PersonalityBench Informativeness indexChat & Writing46.779.1+32.7
Chess Puzzles (task version 2.0)Games79.048.0−31.0
BuseyBench SVGMultimodal7.24.1−30.8
AA-PersonalityBench Slop indexChat & Writing6.737.2+30.8
GameBench 2Games99.168.4−30.7
MysteryMechanismScience51.821.2−30.6
Conceptual Reasoning IndexIndexes79.958.4−29.5
LMArena Code Arena WebDevCode18361621−28.2
FrontierMath Tiers 1-3 (v2) - PrivateMath95.167.4−27.7
BrokenArXiv 09/2026 (Native Harnesses)Math80.554.7−25.8
LMCAGeneral QA70.647.0−23.6
CritPt (avg@5)Science35.714.3−21.4
ACCoRDGeneral QA89.468.1−21.3
AA-PersonalityBench Conviction indexChat & Writing75.396.2+21.2
AA-PersonalityBench Transparency indexChat & Writing81.060.0−21.2
ARC-AGI-2General QA93.872.9−20.8
APEX-Accounting 1.1Finance & Accounting70.450.1−20.3
AA-Briefcase v1.1 EloAgents17851415−20.3
Medical Long Context Reasoning (MLCR-AA, October 2026 methodology)Health & Medicine43.923.9−20.0
AA-PersonalityBench Expressiveness indexChat & Writing81.061.9−19.2
HealthBench ProfessionalHealth & Medicine71.452.9−18.5
τ³-Banking (Mercor run, bm25_grep retrieval)Agents55.737.8−17.9
BioMysteryBench (Vals Terminus 2)Science83.767.8−15.9
SciCode 1.0.1 (Artificial Analysis)Code67.751.9−15.9
SWE-bench Multimodal (Vals v1, 480 Tasks)Code62.346.5−15.8
Epoch Capabilities Index (ECI)Indexes174156−14.4
BioSecBench-Structure V1 (58 evaluations)Science41.427.6−13.8
DAXBench v2Code88.375.5−12.8
FrontierCode 1.1 Main (Artificial Analysis)Code56.244.0−12.2
Ramp Accounting BenchFinance & Accounting52.240.8−11.4
GDPval-AA v2.1Professional Work18311600−10.2
Excel Modeling Benchmark v1.1Finance & Accounting77.467.8−9.6
DTBenchGeneral QA99.289.9−9.3
Vibe Code Bench v1.1Code93.884.7−9.0
Immigration BenchLegal74.065.6−8.3
MineBench — Bradley-Terry RatingGames19321822−6.7
Vals Finance IndexIndexes63.657.1−6.5
Harvey LAB-AA v1.1 (hallucination-gated)Legal7.21.7−5.6
MacroscopeBenchCode76.070.5−5.5
AutomationBench-AA 1.0.6Agents74.268.9−5.3
Tax Agent BenchFinance & Accounting67.062.5−4.5
Finance Agent v2.1Finance & Accounting69.565.4−4.0
ARC-AGI-1 Semi-PrivateGeneral QA98.594.5−4.0
DualEntry Accounting AI BenchmarkFinance & Accounting81.284.2+3.0
HealthBenchHealth & Medicine60.161.9+1.8
DeepSWE v1.1Code74.372.6−1.7
SWE-bench MultilingualCode96.698.2+1.6
OTIS Mock AIME 2024-2025Math99.798.3−1.4
HealthBench (length-adjusted, Artificial Analysis)Health & Medicine60.460.9+0.5

Claude Fable 5.2vsDeepSeek V4.1 Flashnormalised score marginsData Nov 15, 2026Data: Artificial Analysis, Vals AI, Epoch AI and 17 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…