jaredlogan on nostr

Stepping away from the Anthropic subscription fully.

-> Umbrel running OpenwebUI for one seamless chat UI across all models (below)
-> Mac mini running MapleAI Proxy (for Private TEE) + RoutsrD + Local Ollama (failsafe local model)
-> Macbook Air & Mini running Claude Code CLI + OpenCode Go + Zen + PPQ + Routstr

All connected via NostrVPN has been pretty seamless. Even Obsidian Copilot can talk to all of these.

Should be keeping the same full workflows going without losing any model access, but dropping to ~$30/month + PAYG for larger models instead of a flat $100-$200 subscription to big tech.

⚔️ Gotta save that diaper and daycare money

@routstr @Sirius @PayPerQ @Umbrel ☂️ @Maple

151❤️3❤️1👍1💜1💯1🚀1

. · 1w

Happy to see Routstr being useful to you. RoutstrD got an update yesterday and I see all the providers being dynamically chosen so it should be pretty reliable option. Many routstr nodes are currently offering inference at 50% or more off the upstream costs. FYI routstr is also looking for full tim...

ngmi · 1w

I use nano-gpt mostly; can pay in XMR (or btc etc), $12 subscription for chat/API. Also TEE and ZDR models. It's all in one basically

Aedifico · 4d

Does it mean you only use frontier through api cost, or no frontier at all? My dream solution now is running 32B local models, and get them work for private scenarios, and use frontier for only no private data. So POCs, prototypes, frontend, apps with dummy data, and using my own models for the pri...