Vivi — Avatar for ChatGPT

A living avatar for your existing ChatGPT conversations.

Keep using ChatGPT. Add a little presence with Vivi. Vivi places Lumi or Luca, animated avatars, alongside your conversations on chatgpt.com. They listen while you type, think while a response is being prepared, and move their mouths and gesture while a reply is streaming. • Choose between Lumi and Luca, each with their own expressions and poses. • Choose the original hair color or five additional colors for each avatar. Your choices are saved separately. • Animated idle, listening, thinking, and speaking states, with multiple speaking poses and gentle idle motion. • Optional expressive reactions, including happy, sad, angry, surprised, pointing, and heart gestures. • Drag, resize, or hide the avatar and set your own display name. • Preview hair colors while Settings stays beside the avatar. Click outside to close Settings. • Optional model-directed expressions: copy the setup message in Settings and send it once in a text conversation. Vivi reads supported reply tags and hides them visually. Requires ChatGPT to follow the instruction. • Experimental voice reactions use available ChatGPT WebRTC audio-level statistics. Vivi does not record audio or request microphone access. Compatibility may vary. Expression tags are not used in voice mode. Vivi runs locally. It does not send your conversations or audio levels to a developer server. Only settings are saved locally. No separate API key is required. For desktop Chrome on chatgpt.com, not the ChatGPT mobile or desktop app. Website changes can affect detection. Text mouth motion is a visual animation, not phoneme-accurate lip sync. Vivi is independent and not affiliated with or endorsed by OpenAI. ChatGPT is a trademark of OpenAI. Support: ovo0o0828@gmail.com