Pick a model up top and ask anything. Everything runs straight from your browser to the model — nothing routes through a server of mine.
Capabilities
Experimental: this mode runs on a different provider — Lelapa AI's Vulavula API, built for South African and other African languages — not Puter. It needs your own free API key from vulavula.lelapa.ai, and unlike everything else here, that key is called directly from your browser, which may get blocked by the provider's CORS policy since their API wasn't necessarily built for that. Language codes are my best read of their docs, not confirmed against a live response — if a translation fails, that's the first thing to suspect.
Speak in a different voice for this session: upload or record 10–15 seconds of one person talking clearly in a quiet room. It's sent along with each clip and isn't saved anywhere — Lunga's saved voice stays the default.
Max 30 seconds · m4a, mp3, wav, ogg or webm · up to 5 MB