GPT-6 Sol (xHigh): benchmark results

Provider: OpenAI. Access: API.

Unified ELO 1760 ± 1, rank #26 of 3363 rated models, from 59 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
AutomationBench-AA - Sales - Raw Objective Completion90.07Raw Objective Completion (%)100
AA-Omniscience Index - Software Engineering (SWE) - Rust84Omniscience Index98.6
AutomationBench-AA - Operations - Raw Objective Completion96.51Raw Objective Completion (%)98.4
AA-Omniscience Index - Software Engineering (SWE) - HTML80Omniscience Index97
AA-Omniscience Index - Software Engineering (SWE) - JavaScript76.36Omniscience Index96.3
AA CritPt28Accuracy (%)96.2
AA GDP.pdf - Construction - Criterion Pass Rate76.22Criterion Pass Rate (%)96.1
AA GDP.pdf - Real Estate - Criterion Pass Rate77.57Criterion Pass Rate (%)96.1
AA-Omniscience Index - Software Engineering (SWE) - Java43Omniscience Index95.8
AA-Omniscience Index - Software Engineering (SWE) - C81Omniscience Index95.5
Artificial Analysis Intelligence Index44.1Intelligence Index95.5
AA Humanity's Last Exam46.29Accuracy (%)95

Interactive version: theaggregate.ai/model?slug=gpt-6-sol-xhigh · How It Works · Data refreshed daily, snapshot 2026-09-23.