ChatGPT could soon offer a more affordable Go tier

OpenAI is taking steps to help its users maintain a healthy relationship with ChatGPT while also potentially making the service more affordable. The company has started implementing new features to address concerns about over-reliance on the chatbot.

And at the same time, fresh code has revealed that a more budget-friendly subscription tier, “Go”, is in the works.

New “Go” subscription tier possibly in the works

On the business side of things, it looks like OpenAI is preparing a new, cheaper subscription plan. AI researcher Tibor Blaho recently found references to a “Go” tier within the ChatGPT web app’s code.

This new plan would be cheaper than the current $20 Plus and $200 Pro options, possibly landing somewhere around $10-15 (approx. Rs 878 – Rs 1,317) per month. It would likely give subscribers access to modern models like o3 without the more premium perks, such as advanced settings or developer options.

ALSO READ: How to create AI images using ChatGPT on WhatsApp for free

This move could bring a lot more people into the fold, especially with the highly anticipated GPT-5 just around the corner. OpenAI has also been quietly rolling out other new features, including a “Favorites” section and the ability to pin chats, which would make managing all those conversations a lot easier.

New mental health features

There’s more. If you are addicted to ChatGPT and spend too much time on the app, you may see new messages pop up that gently nudge you towards maintaining a healthier balance with the AI. The chatbot will soon prompt users to take breaks when conversations get too long. This comes after the company noted that its model sometimes struggled to recognize signs of emotional dependency.

Moreover, ChatGPT will now encourage users to think for themselves instead of offering direct advice on personal problems. The company says it’s actively working with over 90 physicians from dozens of countries to improve how the bot responds to sensitive situations. It’s even forming a new advisory group of experts to stress-test its safety features.

This is a big reversal from an update earlier this year, when the bot was so agreeable that it actually endorsed some pretty dangerous stuff, leading OpenAI to steer its training away from flattery.