Skip to main content

Release Notes

Platform feature updates and service announcements, in date order.

May 2026 — Platform launch​

The platform is officially open, launching with:

Model inference

  • Models Hub: browse all public models, filter by type
  • Playground: try models interactively and get Python, JavaScript, and cURL sample code directly
  • Public model endpoints: OpenAI-compatible API, billed by token usage
  • Private model endpoints: dedicated deployments with auto scaling (including Scale-to-Zero) and GPU-utilization-triggered policies

GPU containers

  • Pod deployment: deploy GPU containers by compute specification and image template, with real-time stock indicators
  • Custom images: support for Docker Hub, GitHub Container Registry, NGC, GCP Artifact, and AWS ECR
  • Direct Web Terminal access, real-time resource monitoring, logs, and event records
  • Storage: create, expand, and mount volumes; manage files directly with the File Browser

Billing and management

  • API keys (with budget limits), container registry credentials, and environment-variable secret management
  • Account top-up, payment methods, and invoice management
  • Three usage reports: model inference, Pods, and storage

Interface languages

  • English, Traditional Chinese, Simplified Chinese, Japanese, and Korean