OpenAI’s most recent generative Artificial Intelligence (AI) model, GPT-4o, has begun to be implemented in the free version of the ChatGPT tool with “limited access”, thus coexisting with the model that currently powers it, GPT-3.5 .
GPT-4o, with an ‘o’ standing for ‘omini’, natively supports different modalities, that is, it understands and generates a combination of text, audio and image inputs, with a minimum response time of 232 milliseconds.
In its announcement, OpenAI announced that GPT-4o would be implemented “iteratively” and free of charge for users. ChatGPT Plus users. In addition, it is expected that in the coming weeks the new alpha version of the voice mode with GPT-4o will arrive in this same subscription.
While it reported that this model was intended to reach premium subscribers first, the company has introduced a “limited access” to GPT-4o in the free version of ChatGPT in its web version.
Currently, the free version of this tool offers ChatGPT access powered by GPT-3.5 and limited access to advanced data analysis, file uploads, image recognition, web browsing, and custom GPTs.
The most recent model can be chosen from the options menu that appears at the bottom of the toolbar, next to the ‘Copy’, ‘Regenerate’ and ‘Inappropriate response’ buttons. To these They are followed by a fourth button that allows ‘Change model’.
Clicking on it displays the two models currently available in the free version of ChatGPT. While calling GPT-3.5 “the fastest” – and warning that this model does not support attachments – he refers to GPT-4o as a “smart and fast” model.
Unlike paid subscriptions, the free plan comes with a file upload limit and this is indicated by the tool when adding a second document, for example, a PDF, as Europa Press has been able to verify in its web version . This is when ChatGPT invites the user to upgrade their plan to ChatGPT Plus -with access to or “try again later.”
OpenAI has clarified, for its part, that users in the free tier will be able to use its latest AI model by default “with a limit on the number of messages they can send”, which will vary depending on current usage and demand.
With this, it warns that when GPT-4o “is not available, users will revert to GPT-3.5”, that is, the default language model of the content generation tool.
Likewise, it clarifies that with GPT-4o “users also receive limited access to messages using advanced tools”, such as data analysis, file uploads or discovering and using GPTs, among others.