Post

co/core
@cocore.dev

New inference demand just dropped! This feature prefers Qwen2.5-7B-Instruct-4bit / gemma-4-12b / gemma-3-4b-it-qat-4bit in that order so start hosting those and make sure you're updated to the latest version!

Graze Social@graze.social

Miss a day? Now your feed catches you up. Feed Pulse reads the posts your readers actually looked at, and writes a short daily front page for you - all through @cocore.dev's artisanal, farm-to-table AI inference ATProto co-op! Available *today* for all feeds.

3 reposts14 likes

JP
@byjp.me

As an aside; does co/core have a way to see *demand* for models? So we can pick the models the ecosystem wants & needs, without needing to keep an eye out for posts like this?

JP
@byjp.me

I'm struggling to get a given model running; the web UI's "manage models" and the co/core app never seem to line up (or update between them). I also can't see some of the models you suggest in the co/core app (eg. gemma-4-12b isn't listed in the app) It's all a bit confusing 😅 Any tips?

JP
@byjp.me

(Currently the Models tab in the app says mlx-community/Qwen2.5-7B-In struct-4bit is active, but the Status tab says it's not downloaded. Also the confidential tier says it's applying—but it's been this way for a long time on a 1Gbit connection, nothing seems to be downloading or changing)