Atlas

Compare Models

Grok 4.6 vs Inkling 2 Small

Inkling 2 Small leads by 0.1 capability points (≈95% interval −1.9 to +2.1): statistically tied · P(higher) ≈ 55%

Grok 4.6 wins 27 of 48 shared benchmarks (20 lost, 1 tied)

Capability vs. relative cost
Grok 4.6Inkling 2 SmallClosed 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 4.6vsInkling 2 Smallcapability vs. relative costData Nov 15, 2026Village LLM Atlasatlas.village.dev

Benchmark wins27 – 20 · 48 shared, 1 tied

Comparing matching setups only (48 benchmarks excluded) · suites counted once
← Grok 4.6Inkling 2 Small →

Grok 4.6vsInkling 2 Smallwins by categoryData Nov 15, 2026Data: Vals AI, Artificial Analysis, Mercor and 9 othersVillage LLM Atlasatlas.village.dev

Margins

Each benchmark scaled 0–100% (its declared range, else the field’s range); Δ is Inkling 2 Small minus Grok 4.6.

Legal Research Bench ImmigrationLegal54.527.3−27.3
APEX-Agents 1.1Professional Work77.360.9−16.4
ChartographyMultimodal20.536.7+16.2
Legal Research Bench Civil LitigationLegal37.051.9+14.8
ARC-AGI-2General QA67.181.7+14.6
CrossSearch (pass^3)Search35.023.0−12.0
SWE-bench MultilingualCode92.781.2−11.5
AA-AnalystAgent (pass^5)Professional Work41.351.3+10.0
ARC-AGI-1 Semi-PrivateGeneral QA87.596.5+9.0
Terminal-Bench Science 0.1 (Vals)Science11.42.9−8.6
Arena Code WebDev — HTMLCode16001660+8.4
Terminal-Bench 4.0 (Artificial Analysis)Code21.213.1−8.1
ComplexConstraintsChat & Writing36.544.5+8.0
GDP.pdf (Artificial Analysis, All-pass)Professional Work17.825.2+7.4
Humanity's Last Exam DiamondGeneral QA21.528.7+7.2
Terminal-Bench 2.1Code88.481.3−7.1
SWE-Marathon v1.1Code31.938.8+6.9
DAYJOB: Finance (October 2026 rescoring)Finance & Accounting21.014.3−6.7
CritPt v2 (agentic, avg@5)Science62.355.7−6.6
EnterpriseBench: CoreCraftProfessional Work65.671.8+6.2
FrontierCode 2.0 (Artificial Analysis)Code19.825.4+5.6
Tax Agent BenchFinance & Accounting71.465.8−5.6
MysteryMechanismScience30.625.2−5.4
Vals Science IndexIndexes38.132.7−5.4
DeepSWE v1.1Code66.772.1+5.4
Vibe Code Bench v1.1Code76.281.5+5.3
Legal Research Bench CriminalLegal42.136.8−5.3
CyberBench v1.1 OverallCybersecurity67.772.7+5.1
Terminal-Bench 4.0Code17.212.6−4.5
AA-Omniscience v1.1 IndexIndexes24.916.1−4.4
SciCode 1.0.1 (Artificial Analysis)Code56.552.3−4.2
Arena Code WebDev — ReactCode16311659+3.7
Legal Research Bench HealthLegal73.370.0−3.3
Legal Research Bench Constitutional/Civil RightsLegal36.839.5+2.6
SWE-bench VerifiedCode84.682.2−2.4
Legal Research Bench Business/CommercialLegal48.846.5−2.3
Medical Long Context Reasoning (MLCR-AA, October 2026 methodology)Health & Medicine18.320.6+2.2
BioMysteryBench (Vals Terminus 2)Science72.270.0−2.2
APEX-Accounting 1.1Finance & Accounting60.362.0+1.7
EnterpriseOps-Gym-AAAgents48.346.8−1.5
Legal Research Bench Administrative/RegulatoryLegal62.160.6−1.5
ATLAS-Finance v1.1Finance & Accounting4.76.0+1.3
GDPval-AA v2.1Professional Work16581630−1.2
AA-Briefcase v1.1 EloAgents15461527−1.0
ProgramBenchCode1.00.5−0.5
Terminal-Bench-Science 0.1Science6.25.7−0.5
Legal Research Bench Weighted PassLegal88.387.9−0.5
Legal Research Bench FamilyLegal22.722.70.0

Grok 4.6vsInkling 2 Smallnormalised score marginsData Nov 15, 2026Data: Vals AI, Artificial Analysis, Mercor and 9 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…