Atlas

Compare Models

Gemini 3.8 Flash vs Muse Wonder Full

Muse Wonder Full leads by 4.0 capability points (≈95% interval +1.6 to +6.4): a clear lead over Gemini 3.8 Flash · P(higher) ≈ 100%

Wins 62 of 75 shared benchmarks (13 lost)

Capability vs. relative cost
Gemini 3.8 FlashMuse Wonder FullClosed 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.

Gemini 3.8 FlashvsMuse Wonder Fullcapability vs. relative costData Nov 15, 2026Village LLM Atlasatlas.village.dev

Benchmark wins13 – 62 · 75 shared

Comparing matching setups only (2 benchmarks excluded) · suites counted once
← Gemini 3.8 FlashMuse Wonder Full →

Gemini 3.8 FlashvsMuse Wonder Fullwins by categoryData Nov 15, 2026Data: Vals AI, Artificial Analysis, Arena and 6 othersVillage LLM Atlasatlas.village.dev

Margins

Each benchmark scaled 0–100% (its declared range, else the field’s range); Δ is Muse Wonder Full minus Gemini 3.8 Flash.

ProofBench v1.1Math48.094.0+46.0
Vending-Bench 2Games509412292+43.6
Terminal-Bench 4.0Code19.758.1+38.4
CyberBench v1.1 OverallCybersecurity43.873.6+29.9
AA-Briefcase v1.1 EloAgents12041729+28.8
MASKSafety58.687.3+28.7
Terminal-Bench Science 0.1 (Vals)Science8.637.1+28.6
Terminal-Bench-Science 0.1Science10.537.1+26.7
Artificial Analysis Coding Agent Index v1.5Indexes41.968.0+26.2
Artificial Analysis Coding Agent Index v2.0Indexes33.358.9+25.6
Code MigrationCode36.561.0+24.5
Code Migration v1.1Code38.662.8+24.1
IOI (Vals v2)Code56.979.8+22.9
Bug Hunt BenchCode17.140.0+22.9
Humanity's Last Exam DiamondGeneral QA29.148.5+19.4
Artificial Analysis Intelligence Index v5.0.1Indexes30.850.1+19.2
Artificial Analysis Intelligence Index v5.0Indexes30.549.7+19.2
FrontierCode 2.0 (Artificial Analysis)Code18.737.8+19.0
LMArena Code Arena WebDevCode15831723+18.4
GDPval-AA v2.1Professional Work14421857+18.3
Artificial Analysis Intelligence Index v4.3.2Indexes40.955.8+14.9
CritPt (avg@5)Science18.332.9+14.6
Legal Research BenchLegal38.953.4+14.4
Finance Agent v2 - Market AnalysisFinance & Accounting78.564.9−13.6
Vals Science IndexIndexes35.848.5+12.8
LMArena Agent Arena Net Improvement (Steering Burden)Agents2.39.0+12.6
Harvey's Legal Agent BenchmarkLegal10.021.7+11.7
Humanity's Last Exam (Text-Only)General QA42.153.6+11.5
HLAB-Vals-1.1Legal12.123.3+11.2
BioMysteryBench (Vals Terminus 2)Science62.273.3+11.1
Vals Index v2.2Indexes56.066.8+10.8
AutomationBench-AA 1.0.6Agents60.971.3+10.4
CritPt v2 (agentic, avg@5)Science65.175.1+10.0
GDP.pdf (Artificial Analysis, All-pass)Professional Work22.832.6+9.8
Vals Index v2.1Indexes54.964.6+9.7
Finance Agent v2 - Financial ModelingFinance & Accounting32.623.7−8.9
Financial Coaching BenchFinance & Accounting66.375.0+8.7
Public Benefits Bench v1.1Professional Work65.373.7+8.5
Tax Agent BenchFinance & Accounting66.875.0+8.2
Vals Legal IndexIndexes69.877.9+8.1
Artificial Analysis Progress Index v1Indexes320343+8.1
Immigration BenchLegal63.471.2+7.8
FrontierCode 1.1 Main (Artificial Analysis)Code37.144.9+7.8
Access to Justice BenchmarkLegal67.174.7+7.6
Vibe Code Bench v1.1Code78.786.0+7.3
Medical Long Context Reasoning (MLCR-AA, October 2026 methodology)Health & Medicine15.622.2+6.7
SWE-bench Multimodal (Vals v1, 480 Tasks)Code40.847.3+6.5
RiemannBench (November 2026 rescoring)Math56.863.2+6.4
Finance Agent v2 - General Qualitative AnalysisFinance & Accounting68.174.2+6.1
SciCode 1.0.1 (Artificial Analysis)Code56.662.6+6.0
LMArena Agent Arena Steering BurdenAgents-1.39.6+5.4
LMArena Agent Arena Praise vs ComplaintAgents4.915.4+5.3
Finance Agent v2 - ComparablesFinance & Accounting52.046.8−5.2
DeepSWE v1.1Code73.879.0+5.2
LMArena Agent Arena Confirmed SuccessAgents1.311.1+4.9
MedScribeHealth & Medicine84.589.3+4.8
Vals RSI Index v1.1 (four tasks)Agents19.924.4+4.5
Vals Finance IndexIndexes60.564.6+4.1
Finance Agent v2 - Disclosure AnalysisFinance & Accounting68.571.6+3.1
Finance Agent v2 - Earnings AnalysisFinance & Accounting81.478.6−2.8
Excel Modeling Benchmark v1.1Finance & Accounting72.975.0+2.1
Finance Agent v2 - AdjustmentsFinance & Accounting54.856.7+1.8
Finance Agent v2.1Finance & Accounting65.867.6+1.8
MedCodeHealth & Medicine48.149.9+1.8
MysteryMechanismScience36.535.1−1.4
Finance Agent v2 - PrecedentsFinance & Accounting36.135.0−1.1
LMArena Agent Arena Bash ReliabilityAgents9.98.3−0.8
Finance Agent v2 - General Quantitative AnalysisFinance & Accounting80.980.1−0.8
Finance Agent v2 - All PassFinance & Accounting49.749.0−0.7
ProgramBenchCode1.00.5−0.5
LegalBenchLegal87.087.4+0.4
AA-LCR v1.1Search84.083.7−0.3
AA-Omniscience IndexIndexes29.630.1+0.3
Excel Modeling BenchmarkFinance & Accounting72.271.9−0.3
LMArena Agent Arena Tool HallucinationAgents0.40.0−0.2

Gemini 3.8 FlashvsMuse Wonder Fullnormalised score marginsData Nov 15, 2026Data: Vals AI, Artificial Analysis, Arena 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…