Atlas

Compare Models

Kimi K3.1 Pro vs Muse Ember

Kimi K3.1 Pro leads by 1.4 capability points (≈95% interval −1.1 to +3.9): statistically tied · P(higher) ≈ 87%

Wins 18 of 35 shared benchmarks (16 lost, 1 tied)

Capability vs. relative cost
Kimi K3.1 ProMuse EmberClosed 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.

Kimi K3.1 ProvsMuse Embercapability vs. relative costData Nov 17, 2026Village LLM Atlasatlas.village.dev

Benchmark wins18 – 16 · 35 shared, 1 tied

Comparing matching setups only (32 benchmarks excluded) · suites counted once
← Kimi K3.1 ProMuse Ember →

Kimi K3.1 ProvsMuse Emberwins by categoryData Nov 17, 2026Data: Vals AI, Arena, Artificial Analysis and 6 othersVillage LLM Atlasatlas.village.dev

Margins

Each benchmark scaled 0–100% (its declared range, else the field’s range); Δ is Muse Ember minus Kimi K3.1 Pro.

Legal Research Bench ImmigrationLegal18.245.5+27.3
Terminal-Bench 4.0Code63.639.4−24.2
DAXBench v2Code86.262.8−23.4
Legal Research Bench Business/CommercialLegal27.951.2+23.3
Legal Research Bench CriminalLegal28.150.9+22.8
Legal Research Bench Civil LitigationLegal29.648.1+18.5
BuseyBench SVGMultimodal6.95.4−15.4
Legal Research Bench Weighted PassLegal71.486.2+14.8
Humanity's Last Exam DiamondGeneral QA39.425.1−14.3
Legal Research Bench FamilyLegal13.627.3+13.6
HLAB-Vals-1.1Legal11.725.0+13.3
Code Migration v1.1 (Vals Index v2.2 subset)Code59.747.1−12.6
Legal Research Bench Constitutional/Civil RightsLegal31.642.1+10.5
LMArena Code Arena WebDevCode16861608−10.4
Artificial Analysis Coding Agent Index v2.1Indexes48.639.7−8.9
Harvey LAB (Mercor run, undisclosed ~213-task cohort)Legal13.822.7+8.9
AA-Briefcase v1.1 EloAgents16781520−8.6
Tax Agent BenchFinance & Accounting63.971.4+7.5
τ³-Banking (Mercor run, bm25_grep retrieval)Agents59.853.3−6.5
AutomationBench-AA 1.0.6Agents73.368.1−5.2
Vals Finance IndexIndexes56.060.9+4.9
LMArena Agent Arena Praise vs ComplaintAgents6.6-2.2−4.4
LMArena Agent Arena Net Improvement (Steering Burden)Agents3.91.7−4.2
LMArena Agent Arena Confirmed SuccessAgents2.1-4.6−3.4
Legal Research Bench HealthLegal73.370.0−3.3
LMArena Agent Arena Steering BurdenAgents4.99.8+2.5
Finance Agent v2.1Finance & Accounting61.563.8+2.3
Excel Modeling Benchmark v1.1Finance & Accounting70.072.2+2.2
DeepFinanceFinance & Accounting34.636.5+1.9
GBENCH One-shot Coding GScoreGames58.056.4−1.7
Legal Research Bench Administrative/RegulatoryLegal56.154.5−1.5
Vibe Code Bench v1.1Code84.983.7−1.2
LMArena Agent Arena Bash ReliabilityAgents6.47.6+0.6
APEX-Agents 1.1Professional Work75.975.7−0.2
LMArena Agent Arena Tool HallucinationAgents1.01.00.0

Kimi K3.1 ProvsMuse Embernormalised score marginsData Nov 17, 2026Data: Vals AI, Arena, Artificial Analysis 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…