Atlas

Compare Models

GPT-5.6 Terra vs Inkling 2 Small

GPT-5.6 Terra leads by 0.6 capability points (≈95% interval −1.5 to +2.6): statistically tied · P(higher) ≈ 71%

Inkling 2 Small wins 28 of 45 shared benchmarks (15 lost, 2 tied)

Capability vs. relative cost
GPT-5.6 TerraInkling 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.

GPT-5.6 TerravsInkling 2 Smallcapability vs. relative costData Nov 15, 2026Village LLM Atlasatlas.village.dev

Benchmark wins15 – 28 · 45 shared, 2 tied

Comparing matching setups only (51 benchmarks excluded) · suites counted once
← GPT-5.6 TerraInkling 2 Small →

GPT-5.6 TerravsInkling 2 Smallwins by categoryData Nov 15, 2026Data: Vals AI, Artificial Analysis, Mercor and 7 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 GPT-5.6 Terra.

Terminal-Bench 4.0 (Artificial Analysis)Code35.413.1−22.2
Legal Research Bench Civil LitigationLegal29.651.9+22.2
CritPt v2 (agentic, avg@5)Science69.755.7−14.0
SWE-bench MultilingualCode94.481.2−13.2
AA-Omniscience v1.1 IndexIndexes-9.316.1+12.7
DeepSWE v1.1Code60.272.1+11.9
DAYJOB: Finance (October 2026 rescoring)Finance & Accounting3.314.3+11.0
APEX-Agents 1.1Professional Work71.860.9−10.9
AA-Briefcase v1.1 EloAgents13331527+10.6
Humanity's Sixth SenseMultimodal22.733.3+10.6
Terminal-Bench 4.0Code22.712.6−10.1
Legal Research Bench HealthLegal80.070.0−10.0
Legal Research Bench Business/CommercialLegal37.246.5+9.3
Legal Research Bench ImmigrationLegal18.227.3+9.1
EnterpriseOps-Gym-AAAgents38.546.8+8.3
Terminal-Bench-Science 0.1Science13.85.7−8.1
SRE Bench — Fully SolvedCode9.51.9−7.6
GDPval-AA v2.1Professional Work14611630+7.5
Terminal-Bench Science 0.1 (Vals)Science10.02.9−7.1
Vibe Code Bench v1.1Code74.681.5+7.0
Terminal-Bench 2.1Code88.081.3−6.7
Medical Long Context Reasoning (MLCR-AA, October 2026 methodology)Health & Medicine27.220.6−6.7
Legal Research Bench Constitutional/Civil RightsLegal34.239.5+5.3
Legal Research Bench FamilyLegal18.222.7+4.5
SWE-bench VerifiedCode77.882.2+4.4
Legal Research Bench Administrative/RegulatoryLegal57.660.6+3.0
Legal Research Bench Weighted PassLegal85.187.9+2.8
BrowseCompSearch87.590.2+2.7
ChartographyMultimodal34.036.7+2.7
SciCode 1.0.1 (Artificial Analysis)Code55.052.3−2.7
ARC-AGI-2General QA83.981.7−2.2
APEX-Accounting 1.1Finance & Accounting59.862.0+2.2
SAGE v1.1 - Computer ScienceMath89.291.2+2.0
Legal Research Bench CriminalLegal35.136.8+1.8
FrontierCode 2.0 (Artificial Analysis)Code23.925.4+1.5
GPQA DiamondGeneral QA91.993.0+1.1
SAGE v1.1 - Linear AlgebraMath79.580.5+1.1
SAGE v1.1 - Partial Differential EquationsMath93.094.0+1.0
CyberBench v1.1 OverallCybersecurity72.172.7+0.7
GDP.pdf (Artificial Analysis, All-pass)Professional Work24.625.2+0.6
Tax Agent BenchFinance & Accounting65.265.8+0.6
CharXiv ReasoningMultimodal85.985.6−0.3
Humanity's Last Exam DiamondGeneral QA28.828.7−0.1
ARC-AGI-1 Semi-PrivateGeneral QA96.596.50.0
ProgramBenchCode0.50.50.0

GPT-5.6 TerravsInkling 2 Smallnormalised score marginsData Nov 15, 2026Data: Vals AI, Artificial Analysis, Mercor and 7 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…