Generative AI asstiant

Hi,

I'm planning to test the AI Assistant in my lab, where I'm running Elastic Security with a trial license. I have two basic questions about it:

  1. The documentation states: "You need an account with a third-party generative AI provider, which AI Assistant uses to generate responses. Supported providers are OpenAI, Azure OpenAI Service, and Amazon Bedrock." Do I need a paid subscription like ChatGPT-4, or is a free account sufficient? If a free account is enough, which provider is most popular among Elastic users?

  2. I see there's mention of a local LLM. Does this mean you download it from Hugging Face? If so, which model is recommended?

Thank for your support.

Hi,

Looks like you can Connect to your own local LLM | Elastic Security Solution [8.15] | Elastic

However if you choose chatGPT I dont think there is a free plan for api, but I might be wrong

Hi @arcsons!

The supported (hosted) providers require payment for API access. OpenAI offers a "Pay as you go" option as an alternative to a subscription. At the time of this writing, it's possible to make a one time payment, with a minimum of $5.

Note: It may take a few minutes for pay as you go credits to apply to your account, which means your API key might fail until they are applied.

We have more documentation on connecting to a locall LLM here.

At this time we don't recommend a particular model, but we do provide guidance in the Large language model performance matrix. That said, since you're just getting started, consider trying the assistant with one of the hosted models first if that's an option for you.

This topic was automatically closed 28 days after the last reply. New replies are no longer allowed.