GPT-5.6 Sol vs GPT-6.2 Sol
GPT-6.2 Sol leads by 4.3 capability points (≈95% interval +1.9 to +6.8): a clear lead over GPT-5.6 Sol · P(higher) ≈ 100%
Wins 38 of 48 shared benchmarks (7 lost, 3 tied)
Capability vs. relative cost
GPT-5.6 SolGPT-6.2 SolClosed weightsOpen weights
Benchmark wins7 – 38 · 48 shared, 3 tied
Comparing matching setups only (9 benchmarks excluded) · suites counted once
Area← GPT-5.6 SolGPT-6.2 Sol →
Data: Artificial Analysis, Vals AI, Arena and 16 othersatlas.village.dev
Margins
Each benchmark scaled 0–100% (its declared range, else the field’s range); Δ is GPT-6.2 Sol minus GPT-5.6 Sol.
| ARC-AGI-3 (Semi-Private) | Games | 7.8 | 54.9 | +47.1 |
| Terminal-Bench-Science 0.1 | Science › Multidisciplinary | 22.4 | 67.6 | +45.2 |
| Terminal-Bench Science 0.1 (Vals) | Science › Multidisciplinary | 20.0 | 58.6 | +38.6 |
| Humanity's Last Exam Diamond | Knowledge & reasoning › General knowledge | 30.4 | 64.0 | +33.6 |
| Chartography | Perception & media › Documents & charts | 45.0 | 75.1 | +30.1 |
| Humanity's Sixth Sense | Perception & media › Spatial & visual reasoning | 30.0 | 56.8 | +26.8 |
| Gray Swan IPI Attack Success (k=15) | Safety & alignment | 27.0 | 2.3 | +24.7 |
| Blueprint-Bench 2 | Perception & media › Spatial & visual reasoning | 0.3 | 0.6 | +23.4 |
| Terminal-Bench 4.0 | Software › Software engineering | 37.9 | 60.6 | +22.7 |
| DAYJOB: Finance (October 2026 rescoring) | Professional › Finance, accounting & tax | 11.8 | 32.8 | +21.0 |
| Terminal-Bench 4.0 (Artificial Analysis) | Software › Software engineering | 39.9 | 60.1 | +20.2 |
| LMArena Code Arena WebDev | Software › Frontend & design | 1615 | 1757 | +18.8 |
| FrontierCUA v1 pass@1 | Tool & computer use › Computer & browser use | 7.5 | 25.0 | +17.5 |
| Vals Science Index | Science › Multidisciplinary | 41.5 | 57.8 | +16.3 |
| LMArena Agent Arena Net Improvement (Steering Burden) | Tool & computer use › General agents | 3.7 | 11.3 | +14.9 |
| EnterpriseOps-Gym-AA | Tool & computer use › Customer service & operations | 42.9 | 28.9 | −14.0 |
| DAYJOB: Healthcare (October 2026 rescoring) | Health & medicine | 4.4 | 17.2 | +12.8 |
| AutomationBench 1.0.6 (Zapier) | Tool & computer use › Customer service & operations | 28.8 | 41.3 | +12.5 |
| DeepsecBench — Revalidated V1 (233 issues) | Software › Cybersecurity | 35.3 | 47.3 | +12.0 |
| AutomationBench-AA 1.0.6 | Tool & computer use › Customer service & operations | 60.1 | 71.3 | +11.2 |
| Legal Research Bench | Professional › Legal | 48.1 | 37.0 | −11.1 |
| Bug Hunt Bench | Software › Software engineering | 41.4 | 52.1 | +10.6 |
| Agents' Last Exam v1.1 (ALE v1.1) | Professional › General knowledge work | 30.3 | 40.2 | +9.8 |
| EnterpriseBench: CoreCraft | Tool & computer use › Customer service & operations | 46.2 | 36.4 | −9.8 |
| LMArena Agent Arena Praise vs Complaint | Tool & computer use › General agents | 3.8 | 23.3 | +9.7 |
| AA-Briefcase v1.1 Elo | Professional › General knowledge work | 1479 | 1612 | +7.3 |
| Medical Long Context Reasoning (MLCR-AA, October 2026 methodology) | Health & medicine | 30.0 | 37.2 | +7.2 |
| Vibe Code Bench v1.1 | Software › Frontend & design | 80.5 | 87.6 | +7.1 |
| SWE-sweep | Software › Software engineering | 4.7 | 11.6 | +6.9 |
| BioMysteryBench (Vals Terminus 2) | Science › Life sciences | 71.1 | 77.8 | +6.7 |
| ARC-AGI-2 | Knowledge & reasoning › Abstract reasoning | 92.5 | 85.8 | −6.7 |
| AA-AnalystAgent (pass^5) | Professional › General knowledge work | 47.5 | 53.8 | +6.3 |
| DrugDiscoveryBench | Science › Life sciences | 56.5 | 62.2 | +5.7 |
| Tax Agent Bench | Professional › Finance, accounting & tax | 68.0 | 62.9 | −5.1 |
| LMArena Agent Arena Steering Burden | Tool & computer use › General agents | 3.8 | 13.4 | +4.8 |
| MentalHealthBench | Health & medicine | 46.7 | 50.9 | +4.2 |
| LMArena Agent Arena Confirmed Success | Tool & computer use › General agents | 2.5 | 10.9 | +4.2 |
| TxBench Preclinical Pharmacology V1 Clean Reruns | Science › Life sciences | 61.3 | 65.3 | +4.0 |
| MysteryMechanism | Math | 33.3 | 36.9 | +3.6 |
| RiemannBench (November 2026 rescoring) | Math | 83.2 | 80.0 | −3.2 |
| LMArena Agent Arena Bash Reliability | Tool & computer use › General agents | 10.5 | 4.3 | −3.1 |
| GDP.pdf (Artificial Analysis, All-pass) | Professional › General knowledge work | 27.8 | 30.0 | +2.2 |
| CritPt v2 (agentic, avg@5) | Science › Physical sciences | 77.1 | 78.6 | +1.4 |
| GPQA Diamond | Science › Multidisciplinary | 94.6 | 96.0 | +1.4 |
| HealthBench Professional | Health & medicine | 64.1 | 65.5 | +1.4 |
| LMArena Agent Arena Tool Hallucination | Tool & computer use › Tool calling | 1.2 | 1.2 | 0.0 |
| ARC-AGI-1 Semi-Private | Knowledge & reasoning › Abstract reasoning | 97.5 | 97.5 | 0.0 |
| TaxCalcBench TY25 | Professional › Finance, accounting & tax | 32.0 | 32.0 | 0.0 |
Data: Artificial Analysis, Vals AI, Arena and 16 othersatlas.village.dev