llm + tts

Language model and speech synthesis running side by side in your browser.
No server, no API key. Nothing leaves your machine.

LLM: SmolLM2-1.7B-Instruct Q4_0, single-turn chat via WebGPU with lean.
TTS: Quantized Kyutai Pocket-TTS model, custom Rust inference compiled to WebAssembly.

LLM: Qwen2.5-0.5B-Instruct Q4_0 (~429MB) via lean compiled to WebAssembly.
TTS: Quantized Kyutai Pocket-TTS model, custom Rust inference compiled to WebAssembly.

← trucs.ai