Which API can I use to search for the latest information on the internet? On ChatGPT, I can search using the integrated browser of ChatGPT. Regarding APIs, which API and model can I use? Is it possible to use plugins within an API? Thank you.
I don’t think there’s a ‘latest information on the internet’ API. It’s best if you narrow it down to say, tweets on a topic, or latest posts on a website (reddit, hackernews, forum etc).
Then you could use actions in a custom GPT to fetch data from these sources.
Of course it’s tricky as many of these sources don’t have official APIs or are costly. But there’s ways to make it work by scraping the data, or using a secondary source. For example, you can get Hackernews data here: Hacker News Search powered by Algolia
No API from OpenAI. I think you can wait for Perplexity APIs.
With the API you’d need to utilize function calls or tool calls. There are cookbooks and documentation to show how to do that as well as some packages that can help make that easier.
As to the APIs you’re looking to use it’s unclear if you’re looking for a search engine API or something else. Perhaps a good starting point might be Wikipedia’s (API:Main page - MediaWiki). There are other sources of course…
Have you tried DuckDuckGo api ? It can fetch information from the internet
The GPT-3.5-turbo-instruct model does not provide accurate responses. I have tested it, and it cannot retrieve the latest data at the current moment.
Use LangChain agents and tools and ReACt prompt.
I used DuckGoSearch to scrap the webpages to give latest info
Do you have any guidance on using LangChain agents vs using Assistants with function calling?
For example, I want to get “3 recommended links for further reading” based on a provided topic. It seems I could do that with function calling that either calls web search APIs or something similar.
Are you able to find data for a specific niche? Any help would be appreciated. I’ am trying to find specific data and failing miserably.
You can use Exa search or Tavily search. I find them pretty good for general searches. Fo niche and narrow searches, I used Google search API. However, the problem with the google is that google returns page abstract instead of part of the matching text so you need to scrape the search result to understand whether it is related to you or not. From the other side, Exa and Tavily use LLM to understand your query and to provide better results.