Hosting a local model and not paying thousands per year...

I simply run my output through a frontier model when I need to double check. Leaving me with a few dollars of API calls instead of what was costing me 200 to 250 to get to the same point. With stuff like Qwen 3.8 Flash Next available as an MoE you might be amazed at how close it comes to Frontier. That said workload and task vary. And I will never say I get there as fast, but I do get there far cheaper with the same output.

Btw Havok, i am kind of curious how you look at Nostr these days. Do you use it (orchestration, ngit etc.) inside your AI coding journy as well?

I am not doing that currently. I love Nostr for the social aspect of it to be honest. I have considered doing more with Nostr, but I am a bit concerned doing so may detract from what I find enjoyable about Nostr and cause some frustration I do not currently have with this wonderful protocol and the people that use it. I may have to give it some additional thought. If you have ideas or recommendations I would be more than happy to hear them.

See that is something I had not considered. Now I have more to ponder. I think the repository is a solid idea, and one I will have to dig into. Thank you!

have you ever had contact with @Biz Dunno, i somehow feel you should, you two can bromance about computers and goats together.

Two of my favorite subjects, and who doesn't need another bromance in their life... 😂