Atlas

Compare Models

Gemini 4 Carbon vs Muse Ember

Gemini 4 Carbon leads by 5.5 capability points (≈95% interval +2.9 to +8.2): a clear lead over Muse Ember · P(higher) ≈ 100%

Wins 54 of 61 shared benchmarks (7 lost)

Capability vs. relative cost
Gemini 4 CarbonMuse EmberClosed weightsOpen weights

Benchmark wins54 – 7 · 61 shared

Comparing matching setups only (3 benchmarks excluded) · suites counted once
← Gemini 4 CarbonMuse Ember →
Data: Artificial Analysis, Vals AI, Surge AI and 10 othersatlas.village.dev

Margins

Each benchmark scaled 0–100% (its declared range, else the field’s range); Δ is Muse Ember minus Gemini 4 Carbon.

FoodTruck BenchBusiness simulations7721834639−54.7
Humanity's Last Exam DiamondKnowledge & reasoning › General knowledge69.125.1−44.0
Terminal-Bench-Science 0.1Science › Multidisciplinary62.420.0−42.4
FrontierCUA v1 pass@1Tool & computer use › Computer & browser use42.57.5−35.0
Code Migration v1.1 (Vals Index v2.2 subset)Software › Software engineering80.147.1−32.9
DAXBench v2Software › Data & ML93.662.8−30.9
LMArena Agent Arena Net Improvement (Steering Burden)Tool & computer use › General agents16.11.0−29.5
RiemannBench (November 2026 rescoring)Math80.053.6−26.4
BuseyBench SVGPerception & media › Generation7.95.4−24.6
Artificial Analysis Intelligence Index v5.0.1Composite indexes60.036.0−24.0
Terminal-Bench 4.0Software › Software engineering63.139.4−23.7
HiL-Bench (Human-in-Loop Benchmark)Tool & computer use › General agents50.374.0+23.7
Ramp SWE-Bench v2Software › Software engineering75.451.8−23.6
Artificial Analysis Engineering IndexScience › Engineering65.341.8−23.5
GDP.xlsxProfessional › General knowledge work38.915.4−23.5
Artificial Analysis Legal IndexProfessional › Legal62.038.5−23.5
AA-Omniscience v1.1 IndexKnowledge & reasoning › General knowledge53.57.7−22.9
Terminal-Bench 4.0 (Artificial Analysis)Software › Software engineering62.640.9−21.7
LMArena Code Arena WebDevSoftware › Frontend & design17691608−21.3
LMArena Agent Arena Praise vs ComplaintTool & computer use › General agents39.4-2.7−21.1
HANDBOOK.md Agents v1.1Tool & computer use › Customer service & operations49.829.8−20.0
CritPt v2 (agentic, avg@5)Science › Physical sciences82.362.3−20.0
SimpleBenchKnowledge & reasoning › General knowledge92.672.7−19.9
DrugDiscoveryBenchScience › Life sciences71.552.4−19.1
Artificial Analysis Economics IndexComposite indexes59.641.0−18.7
EnterpriseBench: CoreCraftTool & computer use › Customer service & operations80.562.1−18.4
Artificial Analysis Finance & Accounting IndexProfessional › Finance, accounting & tax62.644.6−18.0
DAYJOB: Healthcare (October 2026 rescoring)Health & medicine35.217.2−18.0
Artificial Analysis Coding Agent Index v2.1Software › Software engineering55.539.7−15.8
Artificial Analysis Healthcare & Medical IndexHealth & medicine56.841.3−15.5
SWE Atlas v1.1 - OverallSoftware › Software engineering71.556.0−15.5
APEX-Agents 1.1Professional › General knowledge work90.475.7−14.7
Artificial Analysis Coding Agent Index v2.0Software › Software engineering59.445.9−13.5
FrontierCode 2.0 (Artificial Analysis)Software › Software engineering40.627.2−13.4
GBENCH One-shot Coding GScoreGames69.056.4−12.6
Artificial Analysis Strategy & Ops IndexProfessional › General knowledge work60.948.7−12.3
ComplexConstraints (November 2026 rescoring)Language & writing › Instruction following70.958.8−12.1
AutomationBench-AA 1.0.6Tool & computer use › Customer service & operations77.868.1−9.7
LMArena Agent Arena Confirmed SuccessTool & computer use › General agents17.2-2.0−9.6
Artificial Analysis Progress Index v1Composite indexes351329−7.7
HLAB-Vals-1.1Professional › Legal17.525.0+7.5
GDP.pdf (November 2026 rescoring)Professional › General knowledge work56.449.2−7.2
GDP.pdf (Artificial Analysis, All-pass)Professional › General knowledge work29.622.8−6.8
Legal Research BenchProfessional › Legal55.849.0−6.7
LMArena Agent Arena Steering BurdenTool & computer use › General agents17.94.8−6.6
Chartography (November 2026 rescoring)Perception & media › Documents & charts85.579.3−6.2
LMArena Agent Arena Bash ReliabilityTool & computer use › General agents-5.76.6+6.2
Vals Finance IndexProfessional › Finance, accounting & tax67.060.9−6.1
Vibe Code Bench v1.1Software › Frontend & design89.183.7−5.4
Tax Agent BenchProfessional › Finance, accounting & tax76.471.4−5.1
AA-Briefcase v1.1 EloProfessional › General knowledge work16111526−4.6
FORTRESS Average Risk Score (ARS)Safety & alignment16.411.9+4.5
Excel Modeling Benchmark v1.1Professional › Finance, accounting & tax76.572.2−4.3
Finance Agent v2.1Professional › Finance, accounting & tax67.263.8−3.3
τ³-Banking (Mercor run, bm25_grep retrieval)Tool & computer use › Customer service & operations55.753.3−2.4
Professional Reasoning Benchmark - LegalProfessional › Legal53.455.6+2.3
Medical Long Context Reasoning (MLCR-AA, October 2026 methodology)Health & medicine19.418.3−1.1
DAYJOB: Finance (October 2026 rescoring)Professional › Finance, accounting & tax23.824.2+0.4
Harvey LAB (Mercor run, undisclosed ~213-task cohort)Professional › Legal23.022.7−0.3
Vals Legal IndexProfessional › Legal75.975.7−0.2
LMArena Agent Arena Tool HallucinationTool & computer use › Tool calling0.80.9+0.1
Data: Artificial Analysis, Vals AI, Surge AI and 10 othersatlas.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…