Atlas

Compare Models

Claude Haiku 5.5 vs Kimi K3.1 Pro

Kimi K3.1 Pro leads by 2.5 capability points (≈95% interval +0.3 to +4.7): a clear lead over Claude Haiku 5.5 · P(higher) ≈ 99%

Wins 55 of 74 shared benchmarks (19 lost)

Capability vs. relative cost
Claude Haiku 5.5Kimi K3.1 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.

Claude Haiku 5.5vsKimi K3.1 Procapability vs. relative costData Nov 15, 2026Village LLM Atlasatlas.village.dev

Benchmark wins19 – 55 · 74 shared

Comparing matching setups only (62 benchmarks excluded) · suites counted once
← Claude Haiku 5.5Kimi K3.1 Pro →

Claude Haiku 5.5vsKimi K3.1 Prowins by categoryData Nov 15, 2026Data: Vals AI, Epoch AI, Arena and 21 othersVillage LLM Atlasatlas.village.dev

Margins

Each benchmark scaled 0–100% (its declared range, else the field’s range); Δ is Kimi K3.1 Pro minus Claude Haiku 5.5.

SimpleQA Verified (Epoch anti-abstention)General QA23.862.2+38.4
AutomationBench-AA 1.0.6Agents36.073.3+37.3
IOI (Vals v2)Code47.381.9+34.6
Terminal-Bench 4.0Code36.463.6+27.3
MineBench — Bradley-Terry RatingGames15821992+24.8
BuseyBench SVGMultimodal4.66.9+23.8
Terminal-Bench Science 0.1 (Vals)Science10.031.4+21.4
SRE Bench — Fully SolvedCode5.727.1+21.4
NoCoT-Bench NCKI (kspine v3)General QA85.9116+21.1
FrontierMath Tier 4 (v2) - PrivateMath46.365.9+19.5
NoCoT-Bench NCRI 15.2General QA68.3103+18.9
Legal Research Bench ImmigrationLegal36.418.2−18.2
τ³-Banking (Mercor run, bm25_grep retrieval)Agents43.059.8+16.8
Legal Research Bench Business/CommercialLegal44.227.9−16.3
DeepSWE v1.1Code59.975.5+15.6
Legal Research Bench CriminalLegal42.128.1−14.0
Bug Hunt BenchCode20.534.3+13.8
Legal Research Bench Weighted PassLegal84.871.4−13.3
Legal Research Bench Constitutional/Civil RightsLegal44.731.6−13.2
LMArena Code Arena WebDevCode15941688+12.3
Design Arena Overall Code CategoriesCode12801353+12.1
Mystery Game PuzzlesGames30.042.0+12.0
Legal Research Bench Civil LitigationLegal40.729.6−11.1
LMArena Text Arena EloChat & Writing14221483+10.6
Chess Puzzles (task version 2.0)Games55.045.0−10.0
DualEntry Accounting AI BenchmarkFinance & Accounting76.286.1+9.9
FrontierCode 2.0 - Pass RateCode31.922.0−9.9
Humanity's Last Exam DiamondGeneral QA29.639.4+9.8
GDPval-AA v2.1Professional Work16291846+9.6
Code Migration v1.1 (Vals Index v2.2 subset)Code50.259.7+9.5
Legal Research Bench FamilyLegal22.713.6−9.1
FrontierCode 2.0 - ScoreCode28.919.9−9.0
MysteryMechanismScience35.126.1−9.0
GameBench 2Games75.184.0+9.0
Harvey LAB (Mercor run, undisclosed ~213-task cohort)Legal4.913.8+8.9
LMArena Agent Arena Net Improvement (Steering Burden)Agents0.54.9+8.4
Furniture Assembly BenchmarkMultimodal47.555.8+8.3
APEX-Agents 1.1Professional Work68.375.9+7.6
HLAB-Vals-1.1Legal4.211.7+7.5
Legal Research Bench HealthLegal66.773.3+6.7
LMCAGeneral QA51.157.6+6.5
Legal Research Bench Administrative/RegulatoryLegal50.056.1+6.1
Vals Science IndexIndexes38.544.4+5.9
AA-Briefcase v1.1 EloAgents15731678+5.8
LMArena Agent Arena Praise vs ComplaintAgents-3.58.0+5.7
Vibe Code Bench v1.1Code90.484.9−5.5
DeepFinanceFinance & Accounting40.034.6−5.4
FrontierML ($200 wallet)Agents34.929.6−5.3
BioMysteryBench (Vals Terminus 2)Science70.475.6+5.2
MMMU ProMultimodal79.184.1+5.0
ALE-BenchCode19962132+4.8
Excel Modeling Benchmark v1.1Finance & Accounting65.570.0+4.5
DAXBench v2Code81.986.2+4.3
ACCoRDGeneral QA81.577.5−4.0
Finance Agent v2.1Finance & Accounting65.461.5−3.9
Conceptual Reasoning IndexIndexes65.367.8+3.5
FrontierMath Tiers 1-3 (v2) - PrivateMath75.178.2+3.2
Political Science LLM BenchmarkProfessional Work65.768.6+2.9
GBENCH One-shot Coding GScoreGames55.658.2+2.6
LMArena Agent Arena Steering BurdenAgents2.46.7+2.2
ProofBench v1.1Math89.091.0+2.0
Tax Agent BenchFinance & Accounting62.263.9+1.7
DTBenchGeneral QA95.093.3−1.7
LMArena Agent Arena Tool HallucinationAgents-2.11.0+1.5
Vals Finance IndexIndexes57.456.0−1.4
Epoch Capabilities Index (ECI)Indexes160162+1.4
OTIS Mock AIME 2024-2025Math98.9100.0+1.1
LiveBench 2.0Agents75.676.7+1.1
MazeBench ASCIIGames1.02.0+1.0
LMArena Agent Arena Confirmed SuccessAgents1.62.7+0.6
Prophet Arena Arena Score (Agentic, OpenCode)Forecasting85.986.1+0.2
APEX-Accounting 1.1Finance & Accounting62.262.4+0.2
LMArena Agent Arena Bash ReliabilityAgents7.47.2−0.1
Prophet Arena Arena Score (Fixed Context)Forecasting85.485.40.0

Claude Haiku 5.5vsKimi K3.1 Pronormalised score marginsData Nov 15, 2026Data: Vals AI, Epoch AI, Arena and 21 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…