AI

OpenAI brings back old ChatGPT models after users throw tantrum

Now, users can access a curated selection of AI models, including the beloved GPT-4o, alongside the latest offerings.

by
Kevin Raposo
August 13, 2025

Image: KnowTechie

Just a heads up, if you buy something through our links, we may get a small share of the sale. It’s one of the ways we keep the lights on here. Click here for more.

OpenAI is giving ChatGPT users a bit of déjà vu: after a storm of complaints, the company has brought back its old model picker, letting folks choose from a menu of past and present AI brains.

According to Engadget, users can now toggle between three flavors of GPT-5—Auto, Fast, and Thinking—plus access GPT-4o, which many fans thought had been tossed in the scrapper.

The original move to yank GPT-4o and other legacy models sparked an outcry, especially since OpenAI did it with zero warning.

As The Verge reports, the backlash was so loud that CEO Sam Altman promised users would get “plenty of notice” if the company ever decides to pull a model again.

Altman made this commitment directly in a post on X (formerly Twitter) that also detailed the new model options and usage limits.

Updates to ChatGPT:

You can now choose between “Auto”, “Fast”, and “Thinking” for GPT-5. Most users will want Auto, but the additional control will be useful for some people.

Rate limits are now 3,000 messages/week with GPT-5 Thinking, and then extra capacity on GPT-5 Thinking…
— Sam Altman (@sama) August 13, 2025

So, how do you actually access these older models? It’s hidden, but not impossible.

How to access ChatGPT’s older models

Just open ChatGPT in your browser, click your username in the bottom left, select Settings, and toggle on “Show legacy models” or “Show additional models.”

Once enabled, you’ll see GPT-4o, 4.1, o3, and the rest in the model picker—right alongside the new GPT-5 variants.

For paid users, ultra-premium GPT-4.5 is still locked behind the $200-a-month Pro tier, because, as Altman bluntly put it, “it costs a lot of GPUs.”

If you’re a settings tinkerer, you’ll also spot GPT-5 Thinking mini and more. But heavy users should note: there’s a 3,000-message-per-week limit on GPT-5 Thinking before you’re bumped to the mini version.

Meanwhile, OpenAI says it’s working to make ‘GPT-5’s personality a little warmer’—without making it as “annoying” as GPT-4o was to some.

OpenAI just did a complete 180 on their model availability after user backlash forced them to bring back the old favorites. Are you team “give us all the options” or do you think too many choices just complicate things?