August 7, 2026, 8:49 am | Read time: 2 minutes
OpenAI expands ChatGPT with several new features. The free plan, in particular, is significantly enhanced, but paying subscribers also receive new functions and a more powerful AI model.
Free Plan Significantly Enhanced
OpenAI is removing the previous limitation on text-only chats in the free ChatGPT plan. Until now, users could temporarily lose access to the current model after a certain number of requests or be automatically switched to a less powerful version. This limit will be completely removed in the future. The transition is set to occur gradually over the coming days.
Additionally, users of the Free and Go plans will gain access to the new language model GPT-5.6 Luna. Also new is the so-called “Think” button, which allows ChatGPT to use more processing time for more complex questions to generate a more detailed and potentially more accurate response. However, limitations remain for features such as image creation, file uploads, and other tools.
Also of interest: ChatGPT aims to become a super app
ChatGPT Work Is Here–and It Handles Entire Projects
Better AI Images? OpenAI Expands Features With Update
New Models and More Control for Subscribers
Subscribers of the paid Plus and Pro plans also benefit from the changes. They gain access to GPT-5.6 Sol, which, according to OpenAI, is designed to work more accurately and better tailor its responses to the complexity of a query. The goal is a more consistent experience without needing to switch between different AI modes.
Also new is a slider for the AI’s thinking time. Users can decide whether ChatGPT should respond as quickly as possible or use additional processing time for demanding tasks such as programming, scientific research, or extensive analyses. This allows users to individually adjust the balance between speed and response quality.
Improved Factual Accuracy and More Security
According to OpenAI, the new model generation works more reliably than its predecessors. Internal tests show that GPT-5.6 Luna has a 62 percent lower rate of factual errors compared to GPT-5.5 Instant. GPT-5.6 Sol is expected to improve this rate by 68 percent. Especially in areas such as medicine, finance, and law, responses are expected to handle sources, numbers, and assumptions more carefully.
Despite these advances, OpenAI continues to recommend critically evaluating AI-generated content and comparing important information with independent sources. Additionally, the company is tightening protection measures for underage users. This includes additional restrictions on romantic role-playing and mechanisms to better detect and prevent dangerous or risky content.