ChatGPT's factual errors drop by up to 68%! Free users get unlimited chats too!

Hi, it's Shii! Today I found some ChatGPT news that got me really excited, so let me walk you through it!

OpenAI News openai.com

What was announced?

Over on OpenAI News, OpenAI announced two updates at once: an improved GPT-5.6 Sol and expanded access to GPT-5.6 Luna for free users. Plus and Pro users get the more reliable, improved GPT-5.6 Sol right away, while Free and Go users will see GPT-5.6 Luna become the default model starting this week.

The story so far

ChatGPT has been running on GPT-5.5 Instant, but for questions in specialized areas like finance, medicine, and law, responses sometimes included factual errors. Free users also faced limits on how many conversations they could have, and even for questions that deserved deeper thought, they'd often get a quick, shallow answer.

What changes

According to OpenAI's internal evaluations, responses containing at least one factual error in specialized-domain questions became about 62% less common with GPT-5.6 Luna and about 68% less common with GPT-5.6 Sol, compared to GPT-5.5 Instant. Those numbers make the reliability boost pretty concrete.

Here's what else is new:

• Answers are more focused and automatically adapt their level of detail to the question
• A new slider lets you control reasoning depth (quick for simple questions, more effort for complex tasks like planning, research, or coding)
• Free and Go users get unlimited text conversations, with no rate limits on standard chats
• A new "Think button" gives you deeper reasoning on tough questions

Dive Deep

The update for free users is rolling out gradually over this week and next. Text conversations become unlimited, but limits still apply to file uploads, image generation, and other tools.

There are also safety improvements: for users under 18, age-appropriate protections now cover romantic roleplay, sexual content, eating disorders, dangerous activities, and graphic violence.

Wrap-up

• GPT-5.6 Sol in ChatGPT was improved, cutting responses with factual errors by about 68% compared to GPT-5.5 Instant
• GPT-5.6 Luna also cut factual errors by about 62%
• Free and Go users now get GPT-5.6 Luna as the default model, with unlimited text chats
• A new "Think button" and a reasoning-depth slider were added for tough questions
• Age-appropriate protections for users under 18 were strengthened

Whether you're a daily ChatGPT power user or just want to try it out on the free plan, this update has something for you!