Introducing Dedicated Container Inference: Delivering 2.6x faster inference for custom AI models

Together AI · 5 months ago

Together AI launched Dedicated Container Inference, enabling teams to deploy custom generative media models like video generation and image processing with built-in autoscaling, queuing, and monitoring. Customers Creatify and Hedra achieved 1.4x to 2.6x inference speedups through the platform's architecture and optimization work from Together's research team. The service allows direct deployment of models trained on Together's GPU Cloud without artifact transfers, reducing operational overhead for teams moving from training to production.