We are using Tavily API as a tool in our AI agent. The search API takes around 40 seconds to respond to queries, but, on retries with the same prompt, is able to come back in < 4 seconds.
Here is a sample prompt - Recent Federal Reserve interest rate hikes impact on U.S. consumer spending and borrowing trends
Here are the parameter we are using on the client -
tavily_client.search(query, search_depth=“advanced”).
Is it possible to get consistent near real time (< 4 sec) responses at all times? We are currently on bootstrap pricing mode, and we intend to move up the tier if the API satisfies our needs.
Thank you for reaching out! A response time of 40 seconds for queries is very unlikely, and I couldn’t find any record of such delays on our end. Typically, response times range between 1 to 10 seconds. You can also check the current status of the API on our status page here: https://status.tavily.com.
Could you share which framework you’re using Tavily as a tool in? Perhaps that is the cause of the delays, and we’d be happy to look into it.
While we’re unable to provide detailed technical information on how results improve over time, I can confirm that as more people research a particular topic, the results do indeed become better.
Let us know if you have any more questions or feedback. We’re here to help and accommodate your needs as you continue exploring the API!
Our stack utilizes LangGraph, where nodes implement function calls using web search tools, including Tavily. Initially, we had set include_raw_content to True with max_results set to 10, which led to 502 errors or long runtimes. We have since disabled raw content and reduced max_results to 5, and we’re now experiencing much faster results.
We will continue to test the API, and so far, we are pleased with the results it is delivering within our stack.
On a side note, our legal team has been trying to reach you regarding enterprise pricing but hasn’t had success. Is there a specific contact we can reach out to in order to discuss your enterprise package?
I’m glad to hear that you’re pleased with the results you’re obtaining using our API within your stack! Setting include_raw_content can sometimes place constraints on our system, but our team is actively working on improving performance. Things should be running more smoothly now if you need to enable it for specific cases.
I apologize for the difficulties your legal team has had in reaching us. You can either fill out this form, and our team will get back to you as soon as possible, or reach out via email at support@tavily.com.
Thank you again for reaching out. Let us know if you have any more questions or feedback!
Thanks for the response.
My team had filled in the form and also reached out to the support email. There was no response. Would you mind getting in touch with this email id for further conversations pls - andy.sershall@informa.com
Thanks
I’ve informed our account managers, Itamar and Lee, about the situation. Lee mentioned that they’ve been trying to reach out as well, with one email sent on August 22nd and another yesterday. Lee has just sent another follow-up email. Please let us know if you haven’t received it or if there’s still an issue with getting in touch. I’d be happy to help resolve this communication gap.