Multimodal inference;
at the speed of light.

Low latency, realtime multimodal model serving.
Starting with speech at 50ms time to first audio → BLOG