There are so many out there, with varying benefits, risks, and ethics. I’d like to know what to recommend when asked, and also what I could use for myself.
Some areas that could be good for discussion:
- Locally hosted models (both for low and high powered devices)
- Open source models (also calling out “open source” models that aren’t actually open source)
- Privacy friendly tools/frontends (ex. DuckDuckGo’s AI chat for anonymous use of some “free” models)
- Unified interfaces for multiple models, or ‘pay as you go’ platforms instead of paying for individual subscriptions
One unified interface I’ve been enjoying is Stability Matrix, which lets you conveniently switch between different image generation frontends without having duplicates of the model files.
Text generation: LM Studio seems the most user friendly IMO
Voice transcription: I’ve been using AllTalk TTS, was a little frustrating to set up but works
I love lm studios interface and how you can search for models but they fucked up their code somehow and it’s about twice as long as any other option when it comes to actually generating the text.