@exolabs So is it using the shared memory to run a model larger than a single model could run? Or are you saying you are running one model x4 on 4 machines? It's a big difference
@theo Yeah I'll be honest this has a guy who has too many Chrome tabs energy... I guess it's fine if this is literally how you track your work. It's definitely not how I track mine...
@theo I just don't like saying chat GPT it doesn't roll off the tongue.
Codex sounds dope
Claude, sounds sophisticated
I feel like it fits the brand, chatGPT should be just the website chat experience imo
@theo There is no "model config" in mine I have saved configs. this is helpful because I switch between frontier and local a ton and I like to be able to have saved configs to providers or even models based on the needs of the model. I'm a dum so if I can do this I'm sure you can =)
@theo I like mine more tbh, but I'm not allowed to open source it!
Things I want out of t3.
More stats, mcp connection control
More vis into instructions/system and custom through the app
Skill control I want know all the static context being given to my agent.