"It delivers more focused answers, adapts its level of detail to the question, avoids unnecessary formatting, and offers a helpful correction when simply agreeing wouldn't be useful," OpenAI said.
What OpenAI changed in GPT-5.6 Sol
OpenAI has pushed an update to its GPT-5.6 Sol model intended for the day-to-day ChatGPT experience. According to the company, Sol was tuned specifically for how people use ChatGPT — not for the longer-running, agentic workflows found in Codex and ChatGPT Work — and now prioritizes focus, adaptable detail, and reduced unnecessary formatting.
OpenAI claims the revisions make Sol less prone to factual errors involving dates, numbers, sources, legal rules, medical information, financial questions, or assumptions. In internal tests cited by the company, responses containing at least one factual error were 68% less common with GPT-5.6 Sol than with GPT-5.5 Instant.
How GPT-5.6 Luna and free access change the product mix
OpenAI is also rolling out GPT-5.6 Luna for Free and Go users. Luna will become the default model for Free and Go accounts this week, and some users are already seeing the change. A Go subscription observed by the reporting cost approximately $10.
Starting next week OpenAI plans to remove the rate limit for text chats on Free and Go plans, although usage will remain subject to abuse protections. The company clarified that limits will still apply to file uploads, image generation, and other ChatGPT tools.
OpenAI reports that a Luna variant reduced responses with at least one factual error by 62% in the same evaluation that produced the Sol figure.
This site is the portfolio.
OSINTSights runs on Cloudflare Workers, D1, R2, and Vectorize, with an AI pipeline on Hetzner ARM. Nubivance designed, built, and operates it. We do the same for clients.
See what we buildThe new reasoning slider and the Think button
BleepingComputer observed a new user control that allows users to adjust the model’s reasoning effort. The slider ranges from "Instant" to "High": Instant responds almost immediately, while High reasoning could take several minutes. OpenAI suggests keeping the slider low for everyday questions and increasing it for tasks such as research, coding, writing, planning, and complex decisions.
Free users will also receive a new Think button that gives GPT-5.6 Luna additional time to work through harder questions. The Think button is presented as the mechanism for people on free plans to request deeper reasoning without switching models or subscribing.
Protections for users believed to be under 18
Alongside capability and access changes, OpenAI is introducing additional safeguards for accounts the company believes belong to users under 18. Those protections include stricter boundaries around romantic roleplay, sexual content, dangerous activities, eating disorders, body-image risks, age-restricted goods, and graphic violence.
OpenAI did not link these content boundaries to any particular model variant in the reporting; the protections were described as an additional layer in the updated system behavior.
What this means for Plus and Pro users, Free/Go users, and enterprises using Codex or Work
- Plus and Pro users: The rollout makes GPT-5.6 Sol the more reliable option inside the regular Chat experience, promising fewer factual errors and more concise answers for simple queries, while offering configurable reasoning for complex tasks.
- Free and Go users: Free accounts gain GPT-5.6 Luna as the default and, next week, will see text-chat rate limits removed, with the Think button available to request deeper reasoning—changes that expand unrestricted conversational access while preserving limits on uploads and image generation.
- Enterprises using Codex or ChatGPT Work: The Sol tuning does not apply to Codex or ChatGPT Work; OpenAI explicitly stated the revised Sol is designed for everyday conversations and will only be available through ChatGPT's regular Chat experience, leaving agentic workflows on a different path for now.
Today's changes are rolling out gradually, with OpenAI emphasizing a balance between broader access and operational protections. The company offers measurable reductions in internal-error rates for both Sol and Luna and new user controls for reasoning depth, while retaining separate product lines for agentic workflows and adding safeguards for under-18 users. How these improvements perform at scale, once rate limits change and more users shift models, will emerge in the coming weeks.
