Atlas

Compare Models

Claude Haiku 5.5 vs Claude Opus 5

Claude Opus 5 leads by 2.8 capability points (≈95% interval +0.5 to +5.0): a clear lead over Claude Haiku 5.5 · P(higher) ≈ 99%

Wins 129 of 174 shared benchmarks (22 lost, 23 tied)

Capability vs. relative cost
Claude Haiku 5.5Claude Opus 5Closed 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 5.5vsClaude Opus 5capability vs. relative costData Nov 15, 2026Village LLM Atlasatlas.village.dev

Benchmark wins22 – 129 · 174 shared, 23 tied

Comparing matching setups only (25 benchmarks excluded) · suites counted once
← Claude Haiku 5.5Claude Opus 5 →

Claude Haiku 5.5vsClaude Opus 5wins by categoryData Nov 15, 2026Data: Vals AI, Mercor, Artificial Analysis and 46 othersVillage LLM Atlasatlas.village.dev

Margins

Each benchmark scaled 0–100% (its declared range, else the field’s range); Δ is Claude Opus 5 minus Claude Haiku 5.5.

Vending-Bench 2Games360211182+45.9
VoxelBench Text RatingGames16962300+43.0
A-Fantasia (Valid Attempts with Retries)Games72.329.7+42.7
Toloka Arena (v3)Agents21.062.0+41.0
IOI (Vals v2)Code47.384.3+37.1
ARC-AGI-2General QA61.390.4+29.2
Mystery Game PuzzlesGames30.059.0+29.0
SWE-bench Verified Extended (Mercor held-out)Code53.382.0+28.7
Extended NYT Connections 940 (Quadratic v1)Games65.794.3+28.6
BrowseComp (Mercor 130-question subset)Search56.384.6+28.3
ARC-AGI-3 (Semi-Private)General QA2.830.2+27.3
FrontierMath Tier 4 (v2) - PrivateMath46.373.2+26.8
MineBench — Bradley-Terry RatingGames15822008+25.8
NoCoT-Bench NCRI 15.2General QA68.3115+25.7
LisanBench Difficulty-WeightedGames27115068+25.2
Roboflow Vision Evals — Detection mAP@75 (September)Multimodal54.330.2−24.1
NoCoT-Bench NCKI (kspine v3)General QA85.9119+23.2
Artificial Analysis Coding Agent Index v1.5Indexes36.859.7+22.9
Roboflow Vision Evals — Detection mAP@50:95 (September)Multimodal53.030.5−22.5
ALE-BenchCode19962622+22.2
DeepResearch Bench II (Mercor 130-task set)Search34.356.1+21.8
Terminal-Bench 4.0Code33.353.9+20.6
AutomationBench-AA 1.0.6Agents36.056.6+20.6
Harvey's Legal Agent Benchmark - Corporate M&ALegal0.020.0+20.0
Harvey's Legal Agent Benchmark - Energy Natural ResourcesLegal0.020.0+20.0
Harvey's Legal Agent Benchmark - Healthcare Life SciencesLegal0.020.0+20.0
Harvey's Legal Agent Benchmark - Structured Finance SecuritizationLegal0.020.0+20.0
TaxCalcBench TY25Finance & Accounting2.020.0+18.0
BuseyBench SVGMultimodal4.66.3+17.4
Terminal-Bench Science 0.1 (Vals)Science10.027.1+17.1
ChessBench CoherenceGames80.763.8−16.9
BrowseComp Extended (Mercor held-out)Search27.544.0+16.5
Terminal-Bench 4.0 (Artificial Analysis)Code32.849.0+16.2
BullshitBench v2 LeaderboardSafety57.073.0+16.0
MedXpertQA MM Extended (Mercor held-out)Health & Medicine48.464.1+15.7
Humanity's Last Exam DiamondGeneral QA29.644.8+15.2
AutomationBench 1.0.6 (Zapier)Agents12.026.9+14.9
Excel Modeling BenchmarkFinance & Accounting59.073.6+14.5
Tax Agent BenchFinance & Accounting62.276.7+14.4
Harvey HLAB — Criteria Pass RateLegal73.487.7+14.3
Wiz Cyber Model Arena v3.9Cybersecurity57.071.2+14.2
Roboflow Vision Evals — Detection mAP@50 (September)Multimodal68.254.4−13.8
Public Benefits Bench v1.1Professional Work63.276.9+13.7
LMCAGeneral QA51.164.5+13.4
Furniture Assembly BenchmarkMultimodal47.560.8+13.3
AA-Omniscience IndexIndexes10.737.1+13.2
Code MigrationCode44.657.5+12.9
Conceptual Reasoning IndexIndexes65.374.7+12.9
LMArena Agent Arena Net Improvement (Steering Burden)Agents0.57.2+12.8
LMArena Code Arena WebDevCode15941691+12.7
SUPERChemScience62.074.7+12.7
ExploitGymCybersecurity9.422.0+12.5
GBENCH One-shot Coding GScoreGames55.668.1+12.5
MedCodeHealth & Medicine51.563.6+12.0
Chess Puzzles (task version 2.0)Games55.043.0−12.0
MedXpertQA MM (200-question set)Health & Medicine67.879.8+12.0
LMArena Text Arena EloChat & Writing14221490+11.8
Humanity's Last Exam (Text-Only)General QA42.754.4+11.7
Legal Research BenchLegal43.855.3+11.5
SM BenchSafety74.385.8+11.5
FrontierMath Tiers 1-3 (v2) - PrivateMath75.185.6+10.5
AA-LCR Extended (Mercor held-out)Search63.873.9+10.1
ProofBench v1.1Math89.099.0+10.0
GameBench 2Games75.184.8+9.7
Vals Science IndexIndexes38.547.9+9.4
FrontierCode 1.1 Extended - Pass RateCode60.769.6+9.0
BioMysteryBench (Vals Terminus 2)Science70.479.3+8.9
APEX-Agents 1.1Professional Work68.377.0+8.7
HealthBench ProfessionalHealth & Medicine64.873.4+8.6
APEX-SWECode55.163.7+8.6
Terminal-Bench-Science 0.1Science20.028.6+8.6
ARC-AGI-1 Semi-PrivateGeneral QA89.097.5+8.5
FrontierSWE V2Code43.852.0+8.2
LisanBench Path LengthGames1289715428+7.9
LMArena Agent Arena Praise vs ComplaintAgents-3.512.2+7.8
Artificial Analysis Intelligence Index v5.0.1Indexes36.544.2+7.7
CritPt v2 (agentic, avg@5)Science64.071.4+7.4
Artificial Analysis Intelligence Index v4.3.2Indexes43.450.8+7.4
FORTRESS Average Risk Score (ARS)Safety9.516.8−7.4
WeirdML v3Science11.719.0+7.3
Eval Connections One-Shot with TrapsGames67.074.0+7.0
Design Arena Overall Code CategoriesCode12801321+6.8
CritPt (avg@5)Science22.629.1+6.6
LiveBench 2.0Agents75.682.1+6.5
Humanity's Last Exam (with tools)General QA57.463.6+6.2
HealthBenchHealth & Medicine61.267.1+5.9
LLM Chess EloGames11431285+5.9
Political Science LLM BenchmarkProfessional Work65.771.4+5.8
FrontierPhysics v0.1Science4.29.5+5.3
Bug Hunt BenchCode20.525.7+5.2
CVDP (517-problem set)Code57.062.2+5.2
DeepSWE v1.1Code59.965.0+5.1
MMMU ProMultimodal79.184.0+4.9
MMLU ProGeneral QA87.492.0+4.6
AA-Briefcase v1.1 EloAgents15731654+4.5
GDPval-AA v2.1Professional Work16291729+4.4
ChessBench AccuracyGames66.971.0+4.4
RCM-Bench-1Health & Medicine88.492.7+4.3
LMArena Agent Arena Confirmed SuccessAgents1.610.2+4.3
DAXBench v2Code81.986.2+4.3
Roboflow Vision Evals — Identification, Strict (September)Multimodal80.284.4+4.2
Roboflow Vision Evals — Identification, Judged (September)Multimodal86.590.6+4.1
Terminal-Bench 2.1Code85.089.1+4.1
KernelBench CUDA Normalized (RTX PRO 6000)Code41.445.5+4.0
APEX-1 (Mercor, mean score)Professional Work64.768.6+3.9
Roboflow Vision Evals — Reasoning, Strict (September)Multimodal72.868.9−3.9
Medical Long Context Reasoning (MLCR-AA)Health & Medicine63.359.4−3.9
AA-LCR v1.1Search78.382.0+3.7
SAGEMath45.849.4+3.6
Harvey LAB-AA v1.0 (criterion pass rate)Legal89.993.5+3.6
Epoch Capabilities Index (ECI)Indexes160164+3.4
RedlineBenchLegal40.644.0+3.4
HardGeoBench Avg DistanceMultimodal13191218+3.1
CharXiv Extended (Mercor held-out)Multimodal82.979.8−3.1
GDPval Extended (Mercor held-out rubric score)Professional Work66.062.9−3.1
ChessBench EloGames875826−2.9
DTBenchGeneral QA95.097.9+2.9
Roboflow Vision Evals — Counting, Judged (September)Multimodal73.070.3−2.7
Roboflow Vision Evals — Counting, Strict (September)Multimodal73.070.3−2.7
Artificial Analysis Progress Index v1Indexes326334+2.7
Blueprint-Bench 2Multimodal0.30.3−2.6
HardGeoBench Country AccuracyMultimodal51.353.8+2.5
HardGeoBench <=10miMultimodal30.032.5+2.5
MMMU-Pro Extended (Mercor held-out)Multimodal31.934.3+2.4
ACCoRDGeneral QA81.583.8+2.3
MysteryMechanismScience35.137.4+2.3
APEX-Accounting 1.1Finance & Accounting62.264.3+2.1
Vibe Code Bench v1.1Code90.488.4−2.0
EBR-bench v4 (card ban, single agent)Games43.845.7+1.9
CursorBench 4.0Code48.446.6−1.8
FrontierCode 2.0 - ScoreCode28.927.2−1.7
LMArena Agent Arena Tool HallucinationAgents-2.10.9+1.5
DistressBenchSafety71.770.2−1.5
MedScribeHealth & Medicine89.591.0+1.4
FrontierCode 2.0 - Pass RateCode31.930.4−1.4
SciCode 1.0.1 (Artificial Analysis)Code55.056.4+1.4
Roboflow Vision Evals (October 2026, five tasks)Multimodal69.670.9+1.3
ProgramBench (Mercor 100-task subset)Code3.04.3+1.3
HardGeoBench <=100miMultimodal45.046.3+1.3
SWE-Marathon v1.1Code50.048.8−1.2
ForecastBench Preliminary Dataset Brier IndexForecasting60.862.0+1.2
LMArena Agent Arena Bash ReliabilityAgents7.49.8+1.2
LMArena Agent Arena Steering BurdenAgents2.44.4+1.0
τ³-Banking (Mercor run, bm25_grep retrieval)Agents43.044.0+1.0
SpeciEval (Pooled English Runs)Safety86.987.9+1.0
GDP.pdf (Artificial Analysis, All-pass)Professional Work20.821.6+0.8
Roboflow Vision Evals — Reasoning, Judged (September)Multimodal74.874.2−0.6
CharXiv (descriptive and reasoning combined)Multimodal94.194.6+0.5
AdvancedIFChat & Writing80.880.5−0.3
Terminal-Bench 2.1 Extended (Mercor held-out)Code37.738.0+0.3
GDPval (Mercor 213-task rubric score)Professional Work85.585.6+0.1
AA-AnalystAgent (pass^5)Professional Work53.853.80.0
AIME 2025Math100.0100.00.0
Harvey's Legal Agent Benchmark - Antitrust CompetitionLegal0.00.00.0
Harvey's Legal Agent Benchmark - Arbitration International Dispute ResolutionLegal0.00.00.0
Harvey's Legal Agent Benchmark - Banking FinanceLegal0.00.00.0
Harvey's Legal Agent Benchmark - Bankruptcy RestructuringLegal0.00.00.0
Harvey's Legal Agent Benchmark - Capital MarketsLegal0.00.00.0
Harvey's Legal Agent Benchmark - Corporate GovernanceLegal0.00.00.0
Harvey's Legal Agent Benchmark - Data Privacy CybersecurityLegal0.00.00.0
Harvey's Legal Agent Benchmark - Emerging Companies Venture CapitalLegal0.00.00.0
Harvey's Legal Agent Benchmark - Employment LaborLegal0.00.00.0
Harvey's Legal Agent Benchmark - Environmental ESGLegal0.00.00.0
Harvey's Legal Agent Benchmark - Funds Asset ManagementLegal0.00.00.0
Harvey's Legal Agent Benchmark - ImmigrationLegal0.00.00.0
Harvey's Legal Agent Benchmark - InsuranceLegal0.00.00.0
Harvey's Legal Agent Benchmark - Intellectual PropertyLegal0.00.00.0
Harvey's Legal Agent Benchmark - International Trade SanctionsLegal0.00.00.0
Harvey's Legal Agent Benchmark - Litigation Dispute ResolutionLegal0.00.00.0
Harvey's Legal Agent Benchmark - Real EstateLegal0.00.00.0
Harvey's Legal Agent Benchmark - TaxLegal0.00.00.0
Harvey's Legal Agent Benchmark - Trusts Estates Private ClientLegal0.00.00.0
Harvey's Legal Agent Benchmark - White Collar Defense InvestigationsLegal0.00.00.0
OTIS Mock AIME 2024-2025Math98.998.90.0

Claude Haiku 5.5vsClaude Opus 5normalised score marginsData Nov 15, 2026Data: Vals AI, Mercor, Artificial Analysis and 46 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…