Atlas

Compare Models

Grok 4.6 vs Kimi K3.1 Pro

Kimi K3.1 Pro leads by 3.8 capability points (≈95% interval +1.7 to +6.0): a clear lead over Grok 4.6 · P(higher) ≈ 100%

Wins 61 of 85 shared benchmarks (21 lost, 3 tied)

Capability vs. relative cost
Grok 4.6Kimi K3.1 ProClosed 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.

Grok 4.6vsKimi K3.1 Procapability vs. relative costData Nov 15, 2026Village LLM Atlasatlas.village.dev

Benchmark wins21 – 61 · 85 shared, 3 tied

Comparing matching setups only (39 benchmarks excluded) · suites counted once
← Grok 4.6Kimi K3.1 Pro →

Grok 4.6vsKimi K3.1 Prowins by categoryData Nov 15, 2026Data: Vals AI, SecureBio, Epoch AI and 23 othersVillage LLM Atlasatlas.village.dev

Margins

Each benchmark scaled 0–100% (its declared range, else the field’s range); Δ is Kimi K3.1 Pro minus Grok 4.6.

Terminal-Bench 4.0Code25.365.7+40.4
ProofBench v1.1Math51.091.0+40.0
Legal Research Bench ImmigrationLegal54.518.2−36.4
IOI (Vals v2)Code47.681.9+34.3
FrontierMath Tier 4 (v2) - PrivateMath31.765.9+34.1
LAB-Bench LitQA2 (SecureBio)Science28.953.1+24.2
BuseyBench SVGMultimodal4.66.9+23.3
ZeroBench (Main questions, pass@5)Multimodal17.039.0+22.0
DeepsecBench — Revalidated V1 (233 issues)Code16.137.5+21.4
Legal Research Bench Business/CommercialLegal48.827.9−20.9
CloudOpsBench v1.0Agents39.719.1−20.6
Terminal-Bench Science 0.1 (Vals)Science11.431.4+20.0
ZeroBench Main Mean pass@1 (five samples)Multimodal11.029.8+18.8
Humanity's Last Exam DiamondGeneral QA21.539.4+17.9
Bio Capabilities Index (SecureBio)Indexes137161+17.7
TCSExamMath40.557.4+16.9
Legal Research Bench Weighted PassLegal88.371.4−16.9
Furniture Assembly BenchmarkMultimodal40.055.8+15.8
GameBench 2Games68.784.0+15.3
micro1 PharmaBenchHealth & Medicine12.527.5+15.0
τ³-Banking (Mercor run, bm25_grep retrieval)Agents45.759.8+14.1
Legal Research Bench CriminalLegal42.128.1−14.0
SimpleQA Verified (Epoch anti-abstention)General QA49.362.2+12.9
τ³-Banking (Fireworks, banking_knowledge)Agents50.763.3+12.6
FrontierMath Tiers 1-3 (v2) - PrivateMath66.078.2+12.3
Chess Puzzles (task version 2.0)Games33.045.0+12.0
ALE-BenchCode18222132+11.0
MineBench — Bradley-Terry RatingGames18111992+11.0
Concept Synth INDUCTION Challenge64 Round 1General QA35.946.9+10.9
Design Arena Overall Code CategoriesCode12881353+10.8
MBCT (SecureBio)Science50.460.7+10.3
LAB-Bench ProtocolQA (SecureBio)Science68.178.1+10.0
LMArena Agent Arena Net Improvement (Steering Burden)Agents-0.24.9+9.6
LMCAGeneral QA48.557.6+9.2
LMArena Code Arena WebDevCode16181688+9.1
Legal Research Bench FamilyLegal22.713.6−9.1
Concept Synth INDUCTION Challenge100 Round 1General QA25.034.0+9.0
NoCoT-Bench NCRI 15.2General QA86.8103+8.8
DeepSWE v1.1Code66.775.5+8.8
Vibe Code Bench v1.1Code76.284.9+8.7
GDPval-AA v2.1Professional Work16581846+8.3
Mystery Game PuzzlesGames34.042.0+8.0
Tax Agent BenchFinance & Accounting71.463.9−7.5
Legal Research Bench Civil LitigationLegal37.029.6−7.4
Conceptual Reasoning IndexIndexes62.567.8+7.3
AA-Briefcase v1.1 EloAgents15461678+7.2
Harvey LAB: In-House Contracting (held-out)Legal10.03.0−7.0
Bug Hunt BenchCode27.334.3+7.0
Harvey LAB held-out set (Fireworks run)Legal22.115.4−6.7
AutomationBench-AA 1.0.6Agents67.073.3+6.3
Vals Science IndexIndexes38.144.4+6.3
Legal Research Bench Administrative/RegulatoryLegal62.156.1−6.1
ACCoRDGeneral QA71.677.5+5.9
Legal Research Bench Constitutional/Civil RightsLegal36.831.6−5.3
LMArena Agent Arena Praise vs ComplaintAgents-2.58.0+5.2
GPQA-Bio (SecureBio)Science87.482.2−5.2
LMArena Text Arena EloChat & Writing14561483+4.6
MysteryMechanismScience30.626.1−4.5
LMArena Agent Arena Steering BurdenAgents-1.86.7+4.3
DAYJOB: Finance (October 2026 rescoring)Finance & Accounting21.017.0−4.0
DTBenchGeneral QA97.393.3−4.0
MMMU ProMultimodal80.584.1+3.6
BioMysteryBench (Vals Terminus 2)Science72.275.6+3.3
Epoch Capabilities Index (ECI)Indexes158162+3.3
NoCoT-Bench NCKI (kspine v3)General QA112116+3.2
GBENCH One-shot Coding GScoreGames55.458.2+2.8
LAB-Bench CloningScenarios (SecureBio)Science94.291.5−2.7
LMArena Agent Arena Confirmed SuccessAgents-2.62.7+2.7
APEX-Accounting 1.1Finance & Accounting60.362.4+2.1
HPCT (SecureBio)Science61.663.5+1.9
Access to Justice BenchmarkLegal71.369.5−1.9
LiveBench 2.0Agents74.976.7+1.8
LMArena Agent Arena Bash ReliabilityAgents10.07.2−1.4
APEX-Agents 1.1Professional Work77.375.9−1.4
WCB (SecureBio)Science50.351.5+1.2
VCT (SecureBio)Science48.349.4+1.1
ORCA-bench Hard RCA (Fireworks 324-task subset)Agents32.132.9+0.8
OTIS Mock AIME 2024-2025Math99.2100.0+0.8
MMLU-Bio (SecureBio)Science91.891.2−0.6
EnterpriseBench: CoreCraftProfessional Work65.666.1+0.5
Prophet Arena Arena Score (Agentic, OpenCode)Forecasting85.686.1+0.5
Prophet Arena Arena Score (Fixed Context)Forecasting85.085.4+0.4
LMArena Agent Arena Tool HallucinationAgents1.01.00.0
DAXBench v2Code86.286.20.0
Legal Research Bench HealthLegal73.373.30.0

Grok 4.6vsKimi K3.1 Pronormalised score marginsData Nov 15, 2026Data: Vals AI, SecureBio, Epoch AI and 23 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…