WHEN will the gpt-4.o mini API support access to live internet information?

API
1

When will the API tools support the topics covered in the below article?

  1. Internet access
  2. Images
  3. Memory
  4. Documents and images
1 Like
2

You can already achieve that locally via tools.

My bot can crawl the internet and use search. You can configure it to use gpt-4o-mini.

2 Likes
3

Of course, but I am asking about availability directly in use of the API.

2 Likes
4

I think you are asking the wrong question, with respect.

These features are generally model agnostic.

Watch what they add to Assistants API over time:

https://platform.openai.com/docs/assistants/tools

It is unlikely something the Completions API will ever handle.

3 Likes
5

Hi @tmanchester and welcome to the community.

I’ve actually used Perplexity’s online Sonar models, and they were quite good - it basically acts as a GPT-4 + Google via a single API call. Maybe try them out while we wait for SearchGPT with API access.

3 Likes
6

Unfortunately waiting for features like these to be added to the API is a bit like waiting for Apple to release a calculator on the iPad rather than just downloading one.

Luckily there’s actually an OpenAI Cookbook recently released that guides you through implementing it yourself!

3 Likes
7

But Apple did just release a calculator app on iPad… :wink:

2 Likes
8

Exactly! Chances are it will come eventually, but there are no promises or indications as to when!

1 Like
9

But it’s not a critical priority.

Just use a local function to implement web search. There are several good API options.

My chatbot has local functions for Serp and Jina.

1 Like