GPT-5.6 Sol (Medium): benchmark results

GPT-5.6 Sol evaluated at the medium reasoning-effort setting. Provider: OpenAI. Released 2026-07-09. Access: API.

Unified ELO 1733 ± 1, rank #49 of 1761 rated models, from 43 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
AA Terminal-Bench Hard62.88Accuracy (%)99.5
AA Omniscience - Science, Engineering & Mathematics54.5Accuracy (%)98
AA Omniscience - Business49.5Accuracy (%)97
AA Omniscience - Humanities & Social Sciences56.3Accuracy (%)96.9
AA-Omniscience Accuracy57.75Accuracy (%)96.5
AA Omniscience - Health50Accuracy (%)96.1
AA Omniscience - Software Engineering (SWE)83.2Accuracy (%)95.7
Wolfram LLM Benchmarking Project67.4Correct Functionality (%)95.1
AA GPQA Diamond92.63Accuracy (%)94.8
SWE-rebench62.34Resolved (%)94.8
Artificial Analysis Intelligence Index45.97Intelligence Index94.6
LLM Chess (Saplin)1240ELO94.4

Interactive version: theaggregate.ai/model?slug=gpt-5-6-sol-medium · How It Works · Data refreshed daily, snapshot 2026-09-05.