OpenAI has rolled out significant upgrades to its popular chatbot, ChatGPT, offering more reliable and accurate responses across all user tiers, including free accounts. The changes focus on making the AI model more direct, factually accurate, and consistent in its answers, with a new slider allowing users to adjust the level of reasoning and intelligence.
One of the key upgrades is GPT-5.6 Sol, designed for everyday conversations and available exclusively through ChatGPT’s regular chat experience. This version has been specifically tuned for how people use ChatGPT, rather than for complex workflows found in other AI applications. As a result, users can expect more focused answers, adapted to the question at hand, without unnecessary background information or formatting. For example, asking simple questions like “Who is the president of the United States?” will yield a direct answer, while more complicated queries, such as “How are elections held in the US?”, will provide sufficient detail with clear recommendations.
OpenAI claims that GPT-5.6 Sol reduces factual mistakes by up to 68% when handling sensitive topics like dates, numbers, sources, legal rules, medical information, financial questions, or assumptions. In contrast, free users and those on lower-tier plans will now have access to GPT-5.6 Luna, which has been shown to reduce factual errors by 62%. This model will become the default for free and Go users this week.
One of the most notable features is a new Think button, available to all users, including those with free accounts. When tapped, this button gives GPT-5.6 Luna additional time to reason through harder questions, providing more accurate answers in situations where deeper reasoning is required.
The update does not affect GPT-5.6 Sol in Codex or Work, which are designed for more complex and specialized tasks. Additionally, OpenAI has introduced new protections for users believed to be under 18, including stricter boundaries around romantic roleplay, sexual content, and other sensitive topics.
It’s worth noting that these changes are rolling out gradually, so not all users will see the updates immediately. However, with a free account, users can test the Think button and experience the benefits of GPT-5.6 Luna firsthand.
As AI-powered chatbots like ChatGPT become increasingly ubiquitous, it’s essential for security teams to stay ahead of potential threats. To that end, OpenAI is continually refining its models to improve accuracy and reliability, while also implementing measures to protect users from sensitive content.
Source: Bleeping Computer — 2026-08-06