GPT-5.5 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%
Wins 31 of 47 shared benchmarks (14 lost, 2 tied)
Capability vs. relative cost
GPT-5.5Inkling 2 SmallClosed weightsOpen weights
Benchmark wins14 – 31 · 47 shared, 2 tied
Comparing matching setups only (7 benchmarks excluded) · suites counted once
Area← GPT-5.5Inkling 2 Small →
Data: Vals AI, Artificial Analysis, Thinking Machines Lab and 13 othersatlas.village.dev
Margins
Each benchmark scaled 0–100% (its declared range, else the field’s range); Δ is Inkling 2 Small minus GPT-5.5.
| JobBench Main | Professional › General knowledge work | 38.3 | 70.5 | +32.2 |
| AA-Briefcase v1.1 Elo | Professional › General knowledge work | 1135 | 1527 | +21.5 |
| EnterpriseBench: CoreCraft | Tool & computer use › Customer service & operations | 51.3 | 71.8 | +20.5 |
| Apex (MathArena) | Math | 80.2 | 63.5 | −16.7 |
| GDPval-AA v2.1 | Professional › General knowledge work | 1357 | 1630 | +12.0 |
| Vibe Code Bench v1.1 | Software › Frontend & design | 69.8 | 81.5 | +11.7 |
| LMArena Text Arena Elo | Language & writing › General assistance | 1476 | 1412 | −11.2 |
| Legal Research Bench Constitutional/Civil Rights | Professional › Legal | 28.9 | 39.5 | +10.5 |
| MCP Atlas April 2026 (all 1,000 tasks) | Tool & computer use › General agents | 75.3 | 85.4 | +10.1 |
| BabyVision with Python | Perception & media › Spatial & visual reasoning | 83.6 | 93.6 | +10.0 |
| APEX-Agents 1.1 | Professional › General knowledge work | 70.3 | 60.9 | −9.4 |
| Legal Research Bench Business/Commercial | Professional › Legal | 37.2 | 46.5 | +9.3 |
| Legal Research Bench Family | Professional › Legal | 13.6 | 22.7 | +9.1 |
| Legal Research Bench Immigration | Professional › Legal | 36.4 | 27.3 | −9.1 |
| PerceptionBench | Perception & media › General visual understanding | 55.8 | 64.7 | +8.9 |
| Medical Long Context Reasoning (MLCR-AA, October 2026 methodology) | Health & medicine | 28.9 | 20.6 | −8.3 |
| DeepSWE v1.1 | Software › Software engineering | 64.4 | 72.1 | +7.7 |
| Legal Research Bench Administrative/Regulatory | Professional › Legal | 53.0 | 60.6 | +7.6 |
| Legal Research Bench Civil Litigation | Professional › Legal | 44.4 | 51.9 | +7.4 |
| BrowseComp | Tool & computer use › Search & research | 84.4 | 90.2 | +5.8 |
| APEX-Accounting 1.1 | Professional › Finance, accounting & tax | 56.3 | 62.0 | +5.7 |
| Chartography | Perception & media › Documents & charts | 31.0 | 36.7 | +5.7 |
| Tax Agent Bench | Professional › Finance, accounting & tax | 60.5 | 65.8 | +5.3 |
| MedCode | Health & medicine | 49.1 | 44.1 | −5.0 |
| ComplexConstraints | Language & writing › Instruction following | 49.5 | 44.5 | −5.0 |
| Legal Research Bench Weighted Pass | Professional › Legal | 84.0 | 87.9 | +3.9 |
| SciCode 1.0.1 (Artificial Analysis) | Science › Multidisciplinary | 56.1 | 52.3 | −3.8 |
| SWE-bench Multilingual | Software › Software engineering | 77.7 | 81.2 | +3.5 |
| SWE-bench Verified | Software › Software engineering | 78.8 | 82.2 | +3.4 |
| ARC-AGI-2 | Knowledge & reasoning › Abstract reasoning | 85.0 | 81.7 | −3.3 |
| Legal Research Bench Health | Professional › Legal | 66.7 | 70.0 | +3.3 |
| MedScribe | Health & medicine | 86.9 | 90.2 | +3.3 |
| DeepSearchQA (F1 score) | Tool & computer use › Search & research | 87.8 | 90.9 | +3.1 |
| Terminal-Bench 2.1 | Software › Software engineering | 84.3 | 81.3 | −3.0 |
| HMMT Feb 2026 | Math | 98.5 | 95.8 | −2.7 |
| GDP.pdf (Artificial Analysis, All-pass) | Professional › General knowledge work | 22.6 | 25.2 | +2.6 |
| ARC-AGI-3 (Semi-Private) | Games | 0.4 | 2.6 | +2.1 |
| SRE Bench — Fully Solved | Software › Cybersecurity | 3.8 | 1.9 | −1.9 |
| Terminal-Bench 4.0 (Artificial Analysis) | Software › Software engineering | 14.6 | 13.1 | −1.5 |
| ARC-AGI-1 Semi-Private | Knowledge & reasoning › Abstract reasoning | 95.0 | 96.5 | +1.5 |
| CharXiv Reasoning | Perception & media › Documents & charts | 84.1 | 85.6 | +1.5 |
| AA-AnalystAgent (pass^5) | Professional › General knowledge work | 50.0 | 51.3 | +1.2 |
| Artificial Analysis Progress Index v1 | Composite indexes | 320 | 318 | −0.9 |
| Humanity's Last Exam Diamond | Knowledge & reasoning › General knowledge | 27.9 | 28.7 | +0.8 |
| EnterpriseOps-Gym-AA | Tool & computer use › Customer service & operations | 46.6 | 46.8 | +0.2 |
| Legal Research Bench Criminal | Professional › Legal | 36.8 | 36.8 | 0.0 |
| ProgramBench | Software › Software engineering | 0.5 | 0.5 | 0.0 |
Data: Vals AI, Artificial Analysis, Thinking Machines Lab and 13 othersatlas.village.dev