Hexgrad

Hexgrad is an AI development group focused on speech synthesis. The organization is best known for creating Kokoro, a lightweight yet capable text-to-speech model that has attracted attention in the open-weight AI community for delivering high-quality voice generation at a remarkably small parameter count.

On Venice, Hexgrad is represented by a single specialized model: Kokoro Text to Speech, an 82-million-parameter TTS model released in March 2025. Despite its compact size, Kokoro is designed to produce natural-sounding speech, making it a efficient option for voice generation tasks.

Hexgrad's approach stands out for its emphasis on efficiency in the speech domain. With Kokoro weighing in at just 82M parameters, the model demonstrates that compelling text-to-speech quality does not necessarily require massive scale — a philosophy that aligns well with the broader push toward lightweight, accessible AI models.