September 16, 2026 · Together AIDeprecation
Deployments scale to zero after an inactivity timeout you set
Deployments can now stop themselves when they go unused. Set an inactivity timeout with --inactive-timeout (the inactiveTimeout field in the management API), and if the deployment serves no inference requests for that many minutes, it scales to zero replicas, releasing its hardware and stopping billing.
Related updates
Public Preview: Azure Container Apps serverless GPUs now support Azure AI Foundry models · Azure
Amazon GameLift Servers now supports automatic scaling to and from zero instances · AWS
Announcing on-demand deployment for custom Meta Llama models in Amazon Bedrock · AWS
Managed Airflow Gen 3 development mode and hibernation GA, upgrade rollbacks added · Google Cloud
Announcing on-demand deployment for custom Amazon Nova models in Amazon Bedrock · AWS