Last week, Landian.News reported that OpenAI had renamed ChatGPT’s web-browsing mode from Browsing Alpha to Default (GPT-3.5) with browsing (Alpha). At the time, OpenAI did not specify which version the web-browsing mode was based on, but it was assumed to be GPT-3.5 due to the absence of query limitations.
However, it now appears that this assumption was incorrect. OpenAI renamed the feature to make way for the recently released Web Browsing (GPT-4) Alpha mode.
So, what is the difference between the GPT-3.5 and GPT-4 web-browsing modes? Essentially, there isn’t a significant difference in terms of browsing functionality. The primary change is that the chatbot interacting with users is now GPT-4 instead of GPT-3.5. The online query content is still based on the browsing mode plugin, but any subsequent responses come from GPT-4.
Interestingly, the GPT-4 web-browsing mode does not mention query limitations, while the non-web-browsing GPT-4 mode still imposes a maximum of 25 questions per three hours. It is unclear whether OpenAI has intentionally excluded limitations for the web-browsing mode to encourage more users to test it, or if this is an oversight.
Note: The web-browsing mode is a plugin feature, which is currently being gradually rolled out. Not all users will have access to this mode yet.