Atlas

Compare Models

GPT-5.6 Sol vs Qwen3.8-Max

GPT-5.6 Sol leads by 3.2 capability points (≈95% interval +0.8 to +5.6): a clear lead over Qwen3.8-Max · P(higher) ≈ 100%

Wins 39 of 66 shared benchmarks (27 lost)

Capability vs. relative cost
GPT-5.6 SolQwen3.8-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-5.6 SolvsQwen3.8-Maxcapability vs. relative costData Nov 15, 2026Village LLM Atlasatlas.village.dev

Benchmark wins39 – 27 · 66 shared

Comparing matching setups only (315 benchmarks excluded) · suites counted once
← GPT-5.6 SolQwen3.8-Max →

GPT-5.6 SolvsQwen3.8-Maxwins by categoryData Nov 15, 2026Data: Alibaba, Artificial Analysis, Conceptual Reasoning Benchmark Authors and 13 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-5.6 Sol.

YC-BenchGames7280012095838+47.7
BabaIsBench Stage 0: The IntroGames100.058.0−42.0
GameBench 2Games80.151.4−28.6
Terminal-Bench 4.0 (Artificial Analysis)Code39.918.7−21.2
DeepsecBench — Revalidated V1Code35.415.5−19.9
DeepsecBench — Revalidated V1 (233 issues)Code35.315.5−19.8
DeepsecBenchCode35.616.5−19.0
BabyVisionMultimodal65.582.0+16.5
DeepSWE v1.1Code73.056.6−16.4
OSWorld 2.0 - reported partial reward, patch and step limit unspecifiedAgents62.646.7−15.9
Conceptual Reasoning IndexIndexes68.759.4−12.9
LMCAGeneral QA58.546.2−12.2
LingoQAMultimodal72.684.8+12.2
AutomationBench-AA 1.0.6Agents60.149.2−10.8
AA-Omniscience IndexIndexes22.03.4−9.3
CritPt (avg@5)Science28.620.0−8.6
SimpleVQAMultimodal66.675.0+8.4
Terminal-Bench 2.1Code89.581.3−8.2
WorldVQA AccuracyMultimodal45.153.2+8.1
ERQAMultimodal70.077.8+7.8
AndroidWorld - release-reported mobile useAgents77.685.3+7.7
GDP.pdf (Artificial Analysis, All-pass)Professional Work27.820.2−7.6
𝜏³-BankingAgents44.351.3+7.0
BuseyBench SVGMultimodal6.15.4−6.9
Artificial Analysis Intelligence Index v4.3.2Indexes47.040.2−6.8
Artificial Analysis Intelligence Index v4.3Indexes47.140.3−6.8
Medical Long Context Reasoning (MLCR-AA)Health & Medicine26.119.4−6.7
AA-Briefcase EloAgents14951388−6.2
SLAKEHealth & Medicine85.190.8+5.7
AA-LCR v1.1Search84.078.3−5.7
OmniDocBench v1.5Multimodal86.792.1+5.4
SciCode 1.0.1 (Artificial Analysis)Code57.853.2−4.5
Epoch Capabilities Index (ECI)Indexes163157−4.5
ACCoRDGeneral QA75.971.4−4.5
CharXiv Reasoning (with tools)Multimodal89.193.5+4.4
MathVision - Qwen corrected annotationsMath90.895.2+4.4
RealWorldQA - release-reported visual perceptionMultimodal83.788.0+4.3
Artificial Analysis Progress Index v1Indexes331320−4.2
ClawEval-MM - average of three attemptsAgents78.974.8−4.1
RecreationBench - Qwen internal application recreationCode47.651.7+4.1
ClawEval-MM - pass@3Agents81.277.2−4.0
PMC-VQAHealth & Medicine62.366.2+3.9
PerceptionBenchMultimodal59.763.5+3.8
DTBenchGeneral QA95.792.0−3.7
Video-MMMUMultimodal85.088.7+3.7
MMStarMultimodal82.585.9+3.4
MLVU (multiple-choice M-Avg)Multimodal87.690.8+3.2
ScreenSpot-ProMultimodal81.384.5+3.2
SCI-AGENT-EXAMScience59.662.7+3.1
OSWorld-VerifiedAgents83.286.1+2.9
Nimble Web ContextSearch80.077.3−2.7
GDPval-AA v2Professional Work15851630+2.2
Continual Learning Exam (Trial 1)Agents27.825.7−2.2
GPQA DiamondGeneral QA94.692.6−2.0
ZeroBench (Main questions, pass@5)Multimodal22.024.0+2.0
MMVUMultimodal81.282.4+1.2
MedXpertQA MMHealth & Medicine81.580.4−1.1
MRCR v2 (8-needle, Context Arena) - 128K AUCGeneral QA97.196.1−1.0
NL2Repo - release-reported repository generation scoreCode56.855.9−0.9
Video-MME (with subtitles)Multimodal89.590.4+0.9
Humanity's Last ExamGeneral QA44.543.6−0.9
MMMU ProMultimodal83.082.3−0.7
Dual Exam Agent IndexIndexes20.720.0−0.7
GDPval-AA v2.1Professional Work16191611−0.3
Agents' Last Exam - release-reported pass rate, subset unspecifiedAgents26.727.0+0.3
MRCR v2 (8-needle, Context Arena) - 128K pointwiseGeneral QA92.492.3−0.1

GPT-5.6 SolvsQwen3.8-Maxnormalised score marginsData Nov 15, 2026Data: Alibaba, Artificial Analysis, Conceptual Reasoning Benchmark Authors and 13 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…