Gradium is a voice AI lab (Paris and Silicon Valley) building audio language models for real-time, low-latency Text-to-Speech, Speech-to-Text, and Speech-to-Speech translation. Founded in 2025 by ex-DeepMind / Meta / Kyutai researchers, it exposes a developer API over REST and WebSocket with instant voice cloning, custom pronunciation dictionaries, and multilingual support (EN, FR, DE, ES, PT).
- Website: https://gradium.ai
- Docs: https://docs.gradium.ai/
- API reference: https://docs.gradium.ai/api-reference/introduction
- OpenAPI: https://docs.gradium.ai/api-reference/openapi.json
- GitHub: https://github.com/gradium-ai
- SDKs: Python (
pip install gradium), Rust (cargo add gradium) - Status: https://gradium.statuspage.io/ · Trust: https://trust.gradium.ai/
Backed by: amplify-partners (plus Eric Schmidt, Xavier Niel; $70M seed).
This profile was enriched by the API Evangelist enrichment pipeline (2026-07-19).