Atlas

Compare Models

GPT-5.5 vs Qwen3.8-Max

GPT-5.5 leads by 0.3 capability points (≈95% interval −2.0 to +2.7): statistically tied · P(higher) ≈ 62%

Wins 24 of 44 shared benchmarks (20 lost)

Capability vs. relative cost
GPT-5.5Qwen3.8-MaxClosed 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.

GPT-5.5vsQwen3.8-Maxcapability vs. relative costData Nov 15, 2026Village LLM Atlasatlas.village.dev

Benchmark wins24 – 20 · 44 shared

Comparing matching setups only (188 benchmarks excluded) · suites counted once
← GPT-5.5Qwen3.8-Max →

GPT-5.5vsQwen3.8-Maxwins by categoryData Nov 15, 2026Data: Artificial Analysis, Arena, Vals AI and 11 othersVillage LLM Atlasatlas.village.dev

Margins

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

BabaIsBench Stage 0: The IntroGames100.058.0−42.0
YC-BenchGames12060452095838+31.1
SkillsBench with SkillsAgents62.642.0−20.5
SkillsBench without SkillsAgents48.730.3−18.4
AA-Briefcase EloAgents11581388+13.4
GDPval-AA v2Professional Work13721630+12.7
𝜏³-BankingAgents39.051.3+12.4
GDPval-AA v2.1Professional Work13571611+11.2
Conceptual Reasoning IndexIndexes67.359.4−11.0
SWE-bench VerifiedCode76.485.6+9.2
ACCoRDGeneral QA80.571.4−9.1
AA-Omniscience IndexIndexes20.53.4−8.6
GameBench 2Games59.951.4−8.5
LMCAGeneral QA54.346.2−8.0
OSWorld-VerifiedAgents78.786.1+7.4
Vibe Code Bench v1.1Code58.264.7+6.5
MRCR v2 (8-needle, Context Arena) - 128K pointwiseGeneral QA85.892.3+6.5
AA-LCR v1.1Search84.378.3−6.0
BuseyBench SVGMultimodal6.05.4−5.8
DeepsecBench — Revalidated V1Code21.115.5−5.6
DeepsecBench — Revalidated V1 (233 issues)Code21.015.5−5.6
CritPt (avg@5)Science25.420.0−5.4
DeepsecBenchCode21.216.5−4.7
MRCR v2 (8-needle, Context Arena) - 128K AUCGeneral QA91.796.1+4.4
Terminal-Bench 4.0 (Artificial Analysis)Code14.618.7+4.0
DTBenchGeneral QA96.092.0−4.0
Vision Arena OverallMultimodal12851299+3.7
Terminal-Bench 2.1Code84.381.3−3.0
SciCode 1.0.1 (Artificial Analysis)Code56.153.2−2.9
LMArena Agent Arena Net ImprovementAgents1.72.5+2.6
GDP.pdf (Artificial Analysis, All-pass)Professional Work22.620.2−2.4
Humanity's Last ExamGeneral QA41.443.6+2.2
Epoch Capabilities Index (ECI)Indexes160157−2.1
AutomationBench-AA 1.0.6Agents47.349.2+1.9
Artificial Analysis Intelligence Index v4.3.2Indexes38.440.2+1.8
Artificial Analysis Intelligence Index v4.3Indexes38.640.3+1.7
Medical Long Context Reasoning (MLCR-AA)Health & Medicine17.819.4+1.7
LMArena Text Arena EloChat & Writing14761484+1.4
Finance Agent v2Finance & Accounting51.850.6−1.2
MMMU ProMultimodal81.282.3+1.1
GPQA DiamondGeneral QA93.692.6−1.0
LMArena Agent Arena Bash RecoveryAgents4.12.8−0.7
Artificial Analysis Progress Index v1Indexes320320−0.1
LMArena Agent Arena SteerabilityAgents2.32.0−0.1

GPT-5.5vsQwen3.8-Maxnormalised score marginsData Nov 15, 2026Data: Artificial Analysis, Arena, Vals AI and 11 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…