GPT-5.6 Sol Reduces Factual Errors by 68% in Internal Evaluations

OpenAI deploys GPT-5.6 Sol for paid tiers, cutting factual errors by 68%. Free users gain unlimited text chat and deep reasoning access via Luna model this week.

Woofun AI reports that OpenAI is upgrading ChatGPT to GPT-5.6 Sol for Plus and Pro subscribers, aiming to enhance response focus and factual precision while minimizing superfluous formatting. The model introduces a slider mechanism allowing users to modulate the depth of reasoning between instant responses and complex analysis.

Internal assessments indicate GPT-5.6 Sol decreases factual inaccuracies by approximately 68% compared to GPT-5.5 Instant in finance, healthcare, and legal queries. GPT-5.6 Luna becomes the default for Free and Go users this week, enabling unlimited text conversations and deep reasoning capabilities starting next week, though file uploads and image features remain restricted. This update excludes Work and Codex applications.