Atlas

Compare Models

Claude Haiku 5.5 vs Qwen3.8 Max (0902)

Claude Haiku 5.5 leads by 1.0 capability points (≈95% interval −1.3 to +3.3): statistically tied · P(higher) ≈ 81%

Wins 28 of 49 shared benchmarks (21 lost)

Capability vs. relative cost
Claude Haiku 5.5Qwen3.8 Max (0902)Closed 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 5.5vsQwen3.8 Max (0902)capability vs. relative costData Nov 15, 2026Village LLM Atlasatlas.village.dev

Benchmark wins28 – 21 · 49 shared

Comparing matching setups only (156 benchmarks excluded) · suites counted once
← Claude Haiku 5.5Qwen3.8 Max (0902) →

Claude Haiku 5.5vsQwen3.8 Max (0902)wins by categoryData Nov 15, 2026Data: Artificial Analysis, Mercor, BuseyBench and 4 othersVillage LLM Atlasatlas.village.dev

Margins

Each benchmark scaled 0–100% (its declared range, else the field’s range); Δ is Qwen3.8 Max (0902) minus Claude Haiku 5.5.

AA-PersonalityBench Slop indexChat & Writing10.573.3+63.5
AA-PersonalityBench Expressiveness indexChat & Writing77.223.8−53.8
Medical Long Context Reasoning (MLCR-AA)Health & Medicine63.320.0−43.3
AA-PersonalityBench Informativeness indexChat & Writing75.339.1−36.5
AA-PersonalityBench Conviction indexChat & Writing67.635.3−32.7
AA-PersonalityBench Transparency indexChat & Writing77.244.8−32.7
AA-PersonalityBench Organization indexChat & Writing41.071.4+30.8
FrontierSWE V2Code43.817.8−26.0
AA-PersonalityBench Warmth indexChat & Writing69.548.6−21.2
AutomationBench-AA 1.0.6Agents36.056.2+20.2
VIALS-AA (pass^3)Multimodal49.131.7−17.4
AA-PersonalityBench Imagination indexChat & Writing63.848.6−15.4
CritPt v2 (agentic, avg@5)Science66.351.4−14.9
Code2Video-AA EloCode11921021−13.9
BuseyBench SVGMultimodal4.65.8+12.3
FrontierCode 2.0 (Artificial Analysis)Code34.923.4−11.5
VIALS-AA (average of 3)Multimodal60.550.9−9.5
AA-AnalystAgent (pass^5)Professional Work53.845.0−8.7
Artificial Analysis Strategy & Ops IndexIndexes39.147.3+8.1
Terminal-Bench-Science 0.1Science20.011.9−8.1
AA-3D v1.0 EloCode998954−6.4
Terminal-Bench 4.0Code33.327.0−6.3
Terminal-Bench 4.0 (Artificial Analysis)Code32.838.9+6.1
Humanity's Last Exam DiamondGeneral QA29.624.0−5.6
CritPt (avg@5)Science22.617.7−4.9
Terminal-Bench 2.1Code85.088.8+3.8
Harvey LAB-AA v1.0 (criterion pass rate)Legal89.993.6+3.7
MMMU ProMultimodal79.182.8+3.7
Epoch Capabilities Index (ECI)Indexes160156−3.6
MKA (Artificial Analysis)Professional Work41.037.4−3.5
Artificial Analysis Healthcare & Medical IndexIndexes37.540.8+3.3
Artificial Analysis Engineering IndexIndexes42.539.3−3.3
SciCode 1.0.1 (Artificial Analysis)Code55.052.1−2.9
Artificial Analysis Intelligence Index v5.0.1Indexes36.533.7−2.8
Artificial Analysis Intelligence Index v5.0Indexes36.233.6−2.7
Artificial Analysis Legal IndexIndexes38.741.0+2.3
FrontierCode 1.1 Main (Artificial Analysis)Code45.543.2−2.3
Artificial Analysis Intelligence Index v4.3.2Indexes43.445.4+2.0
AA-LCR v1.1Search78.380.3+2.0
GDP.pdf (Artificial Analysis, All-pass)Professional Work20.822.8+2.0
GDPval-AA v2.1Professional Work16291674+2.0
AA-Briefcase v1.1 EloAgents15731604+1.7
Artificial Analysis Finance & Accounting IndexIndexes43.645.0+1.5
VIALS-AA (pass@3)Multimodal71.472.7+1.2
DAXBench v2Code81.983.0+1.1
Artificial Analysis Progress Index v1Indexes326324−0.9
AA-Omniscience IndexIndexes10.712.0+0.7
Humanity's Last Exam (Text-Only)General QA42.743.1+0.4
Artificial Analysis Economics IndexIndexes42.742.5−0.2

Claude Haiku 5.5vsQwen3.8 Max (0902)normalised score marginsData Nov 15, 2026Data: Artificial Analysis, Mercor, BuseyBench and 4 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…