hbardigital

Some serious hardware porn. What are the economics of this though? I don’t see a price in the article? I saw someone do some rough math in another thread and it was about an order of magnitude cheaper to run your own server like this.

Also, how practical is it to have a set up like this at home? I imagine this won’t run on a 20 amp circuit or even 50 amp?

Would be cool to get one of these with a group of friends. Wonder how many concurrent requests it could handle with something like K3 (though I think 2 of these would be required for that size model)

Looks really promising! I find the claude mobile app a bit lacking ie. slash commands missing (compacting or clearing etc.) Some feedback and questions:

- When I chat via the shellular app, I don't see the messages appear in my terminal. Is that expected? Using cmux fwiw.

- I'm using a Fable agent but the app is showing Opus. Have I been silently downgraded? Both CC and ccstatusline are showing Fable.

- On the home screen it seems to show my agents as well as sub-agents under active sessions whereas under the agents tab I only see what appears to be the main agents. Would be nice for this to be consistent across the screens as well as having a toggle to show/hide sub-agents.

- Loading the sub-agent doesn't bring up any chat history. Would be cool to see how the subagent is working.

- When I re-opened the app after about an hour of having it closed, it wasn't able to reconnect. I did update the mobile app in between so maybe that caused an issue.

Haven't experienced any errors related to children. Will get the random copyright error though.

The speed is definitely impressive. I'm seeing under 5 seconds per image vs ~30 seconds for base NB2.

I built an app for my kids that generates illustrated stories for them with them as the characters. I wanted to prioritize likeness while still stylizing the illustrations. I tested a bunch of models but none seem to come close to maintaining likeness when stylized. I find the others generate generic looking characters.

I'm excited to incorporate this into the onboarding of my app since I want the users to experience the aha moment as soon as possible and waiting half a minute+ isn't ideal. I'll still be using the main NB2 for the actual illustrations as this lite version still has slight issues with nuance and consistency as others have pointed out.

I'm currently building something in the generative AI space and am struggling with pricing. With your fixed price monthly plans, how do you deal with power users who might be blowing through more than $60/month worth of tokens? Do you eat the cost and hope the margins average out? Or have you optimized enough where that's not really a concern?