Nari Labs
Optimized open-source models behind simple production APIs
Nari Labs provides optimized open-source models behind simple production APIs, along with services for dedicated inference and model training. Their platform aims to solve latency and capacity challenges for voice applications, sitting on the quality-latency and latency-cost Pareto Frontier for both Text-to-Speech and Speech-to-Text.
The company's offerings are built for developers and organizations scaling voice AI agents that require high performance and low responsiveness lag. By delivering optimized models like Qwen3-ASR and Qwen3-TTS, Nari Labs enables users to deploy cost-effective, high-accuracy speech infrastructure.
Use Cases
- Developers building real-time voice AI agents
- Teams optimizing Text-to-Speech latency and accuracy
- Engineers reducing Speech-to-Text Word Error Rate
- Organizations deploying open-source voice models via simple APIs
