Atlas

Compare Models

GPT-5.6 Sol vs Qwen3.8 Max (0902)

GPT-5.6 Sol leads by 2.5 capability points (≈95% interval +0.2 to +4.9): a clear lead over Qwen3.8 Max (0902) · P(higher) ≈ 98%

Wins 24 of 31 shared benchmarks (7 lost)

Capability vs. relative cost
GPT-5.6 SolQwen3.8 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.

GPT-5.6 SolvsQwen3.8 Max (0902)capability vs. relative costData Nov 15, 2026Village LLM Atlasatlas.village.dev

Benchmark wins24 – 7 · 31 shared

Comparing matching setups only (89 benchmarks excluded) · suites counted once
← GPT-5.6 SolQwen3.8 Max (0902) →

GPT-5.6 SolvsQwen3.8 Max (0902)wins by categoryData Nov 15, 2026Data: Artificial Analysis, BuseyBench, Co-Design Bench Authors and 6 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 GPT-5.6 Sol.

CritPt v2 (agentic, avg@5)Science77.151.4−25.7
ITBench-AAAgents56.240.3−16.0
FrontierSWE V2Code32.217.8−14.4
CritPt (avg@5)Science28.617.7−10.9
Terminal-Bench-Science 0.1Science22.411.9−10.5
Terminal-Bench 4.0Code37.327.0−10.3
AA-Briefcase EloAgents14951622+7.4
Humanity's Last Exam DiamondGeneral QA30.424.0−6.4
Medical Long Context Reasoning (MLCR-AA)Health & Medicine26.120.0−6.1
Epoch Capabilities Index (ECI)Indexes163156−5.8
SciCode 1.0.1 (Artificial Analysis)Code57.852.1−5.7
GDPval-AA v2Professional Work15851689+5.1
GDP.pdf (Artificial Analysis, All-pass)Professional Work27.822.8−5.0
AA-Omniscience IndexIndexes22.012.0−5.0
EnterpriseOps-Gym-AAAgents42.947.6+4.7
Humanity's Last Exam (Text-Only)General QA47.343.1−4.2
Co-Design BenchCode61.165.3+4.2
AutomationBench-AA 1.0.6Agents60.156.2−3.9
AA-LCR v1.1Search84.080.3−3.7
𝜏³-BankingAgents44.347.8+3.5
BuseyBench SVGMultimodal6.15.8−3.2
Artificial Analysis Progress Index v1Indexes331324−2.7
AA-AnalystAgent (pass^5)Professional Work47.545.0−2.5
GDPval-AA v2.1Professional Work16191674+2.4
Veracode GenAI Code Security (Fall 2026)Cybersecurity70.068.0−2.0
Artificial Analysis Intelligence Index v4.3Indexes47.145.4−1.6
Artificial Analysis Intelligence Index v4.3.2Indexes47.045.4−1.6
GPQA Diamond (Artificial Analysis avg@5)General QA94.192.8−1.3
Terminal-Bench 4.0 (Artificial Analysis)Code39.938.9−1.0
Terminal-Bench 2.1Code89.588.8−0.7
MMMU ProMultimodal82.782.8+0.1

GPT-5.6 SolvsQwen3.8 Max (0902)normalised score marginsData Nov 15, 2026Data: Artificial Analysis, BuseyBench, Co-Design Bench Authors and 6 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…