OpenAI just dropped a significant update for its billion-strong weekly active ChatGPT user base: unlimited text chat. This change applies to both Free and Go tier users, who will also gain access to a new 'Think' button designed for tackling more complex problems. Under the hood, the default model for these tiers is being upgraded to GPT-5.6 Luna, replacing the older GPT-5.5.
It's important to note that 'unlimited' isn't a blanket term. OpenAI has clarified that existing limits on file uploads, image generation, and voice interactions remain in place. Essentially, the shackles are off for pure text-based conversations, while multimodal tasks will still have their own caps.
New Tools for Free Users
The 'Think' button stands out as a key feature in this update. It empowers users to actively engage a higher level of reasoning intensity when faced with challenging problems, rather than passively waiting for the model to decide. This should prove particularly useful for tasks like coding, complex mathematical reasoning, or in-depth analysis of lengthy texts.
OpenAI's rollout plan for Free and Go users is staggered: the new model arrives this week, followed by the unlimited text chat and 'Think' button next week. This two-step deployment might be a pragmatic move to manage server load, especially with such a massive user base.
Enhanced Control for Paid Subscribers
For Plus and Pro subscribers, the focus shifts to GPT-5.6 Sol. This upgraded model is pitched as being more adept at rapid tasks such as information retrieval, drafting suggestions, planning, and decision-making. OpenAI claims it delivers more concise and robust answers.
Additionally, Plus and Pro users gain a 'Thought Slider', offering manual control over how long the model 'thinks' before generating a response. Users can dial it down for quick answers or crank it up for deeper, more deliberate reasoning. This level of adjustable control is a relatively rare sight in conversational AI products.
- The upgraded GPT-5.6 Sol is rolling out to Plus/Pro users today.
- This is a distinct version specifically for chat, separate from the GPT-5.6 Sol previously used for Codex and Work.
Behind the Numbers: Internal Benchmarks
OpenAI also shared some internal comparison data. Compared to GPT-5.5-Instant, GPT-5.6 Luna reportedly shows a 62% reduction in factual errors, while GPT-5.6 Sol boasts a 68% reduction. While these numbers aren't definitive proof, they certainly point towards a positive trend in accuracy.
This update seems to strike a good balance, offering more power to free users and more granular control to paid subscribers. It's a smart play to keep both segments engaged.
What This Means for Users
The removal of text chat limits is a genuine quality-of-life improvement for anyone who uses ChatGPT as a daily information tool. That nagging feeling of 'running out of turns' will finally be gone. More interestingly, the 'Think' button and 'Thought Slider' put the power of deciding how much the model reasons directly into the user's hands, which is a more profound shift than simply removing usage caps.
Once the features are fully live next week, I'd recommend taking a complex problem you've struggled with and running it through both the standard and 'Think' modes to compare the depth and speed of the answers. Paid users should experiment with the 'Thought Slider' from low to high to see if it makes a noticeable difference in multi-step reasoning tasks.
Of course, maintaining separate limits for files, images, and voice interactions signals that OpenAI is still managing its resources carefully. Unlimited text chat might be opening the front door, but there are still a few other gates to navigate.
Comments
No comments yet
Be the first to comment