I don't know much about docker.
I'm running it on both Windows and Ubuntu Linux. Open source and Pro. I think it's been in since 0.47.0
1 Like
Oh I see, the env variable is MB_IS_METABOT_ENABLED, not MB_IS_METABASE_ENABLED 
Thanks again for putting me on track!
BTW to the moderators - MB_IS_METABOT_ENABLED is not documented 
Sorry, silly mistake. I'll edit my earlier comment.
1 Like
I typically encounter issues with context size when testing this feature. Is there a way to configure this to target the gpt-4-turbo model that supports a larger context size?
Do you if there are any plans to be able to use local AI models in later versions of Metabase? e.g. when you work in an environment and not access to OpenAI is possible but having an own/dedicated OpenAI instance running ? Thanks
This is something that many people requested, so it’s being considered
1 Like
Does this work with Microsoft Azure Open AI too? What do i put in as organisation id?
no, it does not
Reading back through this chain of conversation, I presume it still requires Java 11? Further, I noticed a link to a 0.46.1 Metabase version. We're currently on 0.50.31. Assuming I can get Java 11, will the Metabot work with this later iteration of Metabase, or do we have to go back to the earlier Metabase version as well?
Thanks.