GPT-5.6 Sol (Low) — benchmark results

Provider: OpenAI. Released 2026-07-09. Access: API.

Unified ELO 1969 ± 30, rank #32 of 1744 rated models, from 38 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
AA Omniscience - Software Engineering (SWE) - JavaScript89.09Accuracy (%)99.7
AA Omniscience - Health50.3Accuracy (%)99.5
AA Omniscience - Software Engineering (SWE) - Go84Accuracy (%)99.2
AA Omniscience - Software Engineering (SWE) - Rust86Accuracy (%)99.2
AA Omniscience - Software Engineering (SWE) - C88Accuracy (%)99.1
AA Omniscience - Software Engineering (SWE) - PHP92Accuracy (%)99.1
AA Omniscience - Business49.1Accuracy (%)99
AA-Omniscience Accuracy56.43Accuracy (%)98.6
AA Terminal-Bench Hard60.61Accuracy (%)98.5
AA Omniscience - Humanities & Social Sciences54.5Accuracy (%)98.2
AA Omniscience - Software Engineering (SWE) - HTML84Accuracy (%)98.1
AA Omniscience - Software Engineering (SWE) - Kotlin82Accuracy (%)98.1

Interactive version: aibenchmarks.dev/model?slug=gpt-5-6-sol-low · How the rankings work · Data refreshed daily, snapshot 2026-07-10.