still sending all your info to a black-box api when using an llm? you know it's bad. try this instead: you can run dozens of models right in your browser from Google, Microsoft, Mistral, Meta, Qwen, and Smollm. private and secure. you'll thank me later.