dave works with https://openrouter.ai/ now. you can use *any* ai model to drive dave. just tested with google gemini and it works great! here's the CLI config until we have the UI for selecting the backend provider.

this only works if you compile from source for now until we have a new binary release.

back to working on https://jb55.com/lmzap, (lightning version of openrouter) so that you can use lightning instead of fiat+KYC

nostr:note1az07szq3tr5wjx2nwhsas9lvsyfzuae6rv80drapz7c820wlje8s7ky9ua

Reply to this note

Please Login to reply.

Discussion

That’s exciting.

Stop. You want to say that you basically copy pasting X ideas into damus ? Damn

what ideas am I copying ?

Grok

you can't use any ai backend on grok

also dave is more powerful than grok. most things dave can do grok cannot

Ok, I will answer this

Where you got this idea ?!

also I created the in-network version of nostr:npub1tsgw6pncspg4d5u778hk63s3pls70evs4czfsmx0fzap9xwt203qtkhtk4 before grok had this ability. technically they copied that from me.

Oh, sure! Genius Will

can I point it at an ollama or openapi compatible endpoint on the lan?

yes. I have only been able to get https://ollama.com/hhao/qwen2.5-coder-tools working reliably

but I only have 8gb of vram so thats definitely a large factor. i want a dgx spark!

same. damn, i hadn’t seen the dgx spark before that would be ideal.

if using ollama on your local machine it will work even if you have no internet connection. it's pretty cool!