OpenAI expands ChatGPT access with GPT-5.6 updates for free and paid users

OpenAI is gradually introducing updates to ChatGPT that aim to improve response accuracy, consistency, and control over how much reasoning the service applies to a request.

Subscribers on the Plus and Pro plans are receiving GPT-5.6 Sol, a version designed for everyday ChatGPT conversations rather than extended agentic workloads. OpenAI says the model should provide more focused answers, adjust its level of detail to the question, avoid unnecessary formatting, and challenge incorrect assumptions instead of simply agreeing.

The update also adds an intelligence control that allows users to select a reasoning level ranging from Instant to High. Lower settings are intended for quick questions, while higher levels may take considerably longer and are aimed at activities such as research, programming, writing, planning, and complicated decisions.

According to OpenAI’s internal evaluations, GPT-5.6 Sol produced responses containing at least one factual error 68% less often than GPT-5.5 Instant. The company reported a 62% reduction for GPT-5.6 Luna under the same assessment. OpenAI said the testing covered areas including dates, figures, sources, legal guidance, medical information, financial topics, and inferred assumptions.

Free users receive expanded Luna access

GPT-5.6 Luna is becoming the default model for Free and Go users. OpenAI plans to remove the text-chat rate limit for those plans beginning next week, although normal abuse-prevention controls will remain. Restrictions will continue to apply to features such as file uploads, image generation, and other tools.

Free users will also receive a Think option. Selecting it gives Luna additional time to work through questions that require deeper reasoning. The rollout is phased, so availability may vary between accounts.

These changes apply to ChatGPT’s standard conversation experience and do not alter GPT-5.6 Sol deployments in Codex or ChatGPT Work. OpenAI is separately adding stronger safeguards for users believed to be under 18, including tighter limits involving sexual content, romantic roleplay, dangerous activities, eating disorders, body-image concerns, age-restricted products, and graphic violence.