1) I pay $20/month for Nous model access, and spend an additional $20-40/month to Google for Gemini model API access.

2) My main daily driver is Deepseek 4.1-Flash. For bigger coding projects, Argon. I run a local Qwen 3.6-27B quantized to 4-bit so that it fits comfortably in my 4090's VRAM, using this for any easy things that can be done locally. Also have local ComfyUI + Flux + some LoRAs for image work.

3) I exclusively use Hermes

4) N/A for anything I do

5) Yep, just Hermes, for months now, never going with anything else as I love how it works.

- I'm mainly using all this stuff for local projects. For instance, I recently bought a Flydigi Apex 6 Pro controller, but I game exclusively in CachyOS so I had Hermes write my own custom app for controlling haptics and such. I'm also working on a game in Godot, generating pixel art locally.