OpenAI Updates GPT-5.6 Sol and Gives Free Users Unlimited Chats

OpenAI is rolling out an update to GPT-5.6 Sol in ChatGPT that it says produces responses with at least one factual error roughly 68% less often than GPT-5.5 Instant, based on internal evaluations of financial, medical, and legal prompts. The company announced the changes alongside a significant expansion of free access: starting next week, Free and Go users get unlimited text chats on the new default model, GPT-5.6 Luna, plus a Think button that routes harder questions to deeper reasoning.

The release lands at a moment when OpenAI says 1 billion people turn to ChatGPT every week. For a company whose stated mission is, in its own words, "to ensure that artificial general intelligence benefits all of humanity," the commercial and reputational stakes of factual reliability and free-tier access are high. Rivals including Google and Meta are competing aggressively on both price and reach, and removing text rate limits for free users is a direct move in that contest.

What changes for Plus and Pro users

For paying subscribers, OpenAI has updated GPT-5.6 Sol in Chat to be, as the company puts it, "more reliable with facts and provide more focused answers." A new slider — available on web, mobile, and desktop — lets users choose how much thought ChatGPT puts into each response. OpenAI's framing is straightforward: keep it quick for everyday questions, or move the slider up for planning, research, writing, coding, or decisions that need more thought.

The update also collapses a long-standing seam in the ChatGPT experience. The same GPT-5.6 Sol model now powers both Instant responses and deeper reasoning, which OpenAI says creates "one consistent experience." The company describes the intended effect in concrete terms: when a user moves from Instant to higher effort, "it should feel like the model is taking extra time for a more comprehensive answer—not like you're switching to a different model with its own tone or style."

The reliability numbers

The most specific claim in the announcement concerns factual accuracy. In an internal evaluation of financial, medical, and legal prompts requiring factual detail, responses containing at least one factual error were about 62% less common with GPT-5.6 Luna and 68% less common with GPT-5.6 Sol than with GPT-5.5 Instant. OpenAI attributes the improvement to the model "better using the sources it finds to answer your question," with particular gains on answers that depend on dates, numbers, sources, rules, or assumptions.

Those domains — finance, medicine, law — are exactly where model errors carry real consequences, and where regulators and enterprise buyers scrutinize performance. OpenAI did not publish external benchmark results in the announcement, so the figures rest on its own internal evaluation.

The formatting changes are aimed at a familiar complaint among power users: bloated, over-structured answers. The updated GPT-5.6 Sol, OpenAI says, "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." That last point is a nod to sycophancy, the tendency of chatbots to validate users rather than push back, which OpenAI has publicly committed to reducing. A quick question should now get "a direct answer with the context you need," while multi-step planning or research work gets a fuller response that "keeps the main recommendation clear."

Free users get unlimited text and a Think button

The bigger strategic news is on the free tier. GPT-5.6 Luna becomes the default model for Free and Go users this week. Starting next week, those users get unlimited text chats — a departure from the message caps that have long defined ChatGPT's free experience — plus a new Think button for questions that require deeper reasoning, which gives GPT-5.6 Luna more time to work through an answer.

There are limits. The unlimited provision is "subject to abuse guardrails," and limits will still apply to file uploads, images, and other tools. The Think button interface, shown in OpenAI's announcement, also includes an option to upgrade to Plus — the free expansion doubles as a funnel to paid subscriptions.

OpenAI frames the move in expansive terms. "This is a concrete step toward more abundant intelligence: making our latest models more widely available, improving the usefulness and reliability of the answers people get, and letting free users keep text chats going without a rate limit," the company writes. "Access shapes opportunity, and this update gives more people the ability to keep asking, develop an idea, and get help when they need it."

Scope limits and what isn't changing

The updated GPT-5.6 Sol is optimized for everyday chats, and OpenAI says it will only be available in the Chat experience within ChatGPT. The version of GPT-5.6 Sol that powers Work and Codex is not changing as part of this release — a distinction that matters to developers and business users whose workflows depend on predictable model behavior.

Safety measures for users under 18

The accompanying system card, published on OpenAI's deployment safety site, outlines additional measures for users the company believes are under 18. OpenAI trained the model to avoid romantic roleplay, age-restricted challenges, and presenting itself as a substitute for real-world relationships. The company also applied age-appropriate boundaries around sexual content, eating disorders and body-image risks, age-restricted goods, dangerous activities, and graphic violence. The model is additionally trained to encourage connection with trusted people when a teen may need support.

OpenAI says it reinforced this training with system-level protections and has added new evaluations for how its models perform for users under 18, with continued improvement planned in that area. The focus on minors reflects intensifying scrutiny from lawmakers and regulators in the United States and European Union over how AI products interact with children and teenagers.

Why it matters

Two threads run through this release. The first is convergence: merging Instant and Thinking into a single model with an adjustable effort slider removes a product distinction that had existed since OpenAI introduced reasoning models, and mirrors similar controllable-reasoning features competitors have shipped. The second is reach: unlimited free text chats put ChatGPT's latest models in front of the largest possible audience at a time when the marginal cost of serving text queries continues to fall — and when OpenAI's business increasingly depends on that billion-user weekly base justifying its infrastructure investments.

Plus and Pro users can access the updated GPT-5.6 Sol and the new slider starting today. Free and Go users see GPT-5.6 Luna as their default this week, with unlimited text chats and the Think button arriving next week. How the factual-accuracy gains hold up outside OpenAI's internal evaluations — and whether unlimited free access proves economically sustainable — will shape the company's next moves.