Announcing OpenAI-compatible API support for Amazon SageMaker AI endpoints

This article announces OpenAI-compatible API support for Amazon SageMaker AI real-time inference endpoints, enabling seamless integration with existing OpenAI SDKs and frameworks.

• SageMaker endpoints now expose `/openai/v1` path for Chat Completions requests
• Use OpenAI SDK, LangChain, or Strands Agents by changing only the endpoint URL
• Bearer token authentication with time-limited tokens (up to 12 hours) from AWS credentials
• Support for single-model endpoints and multi-model deployments via inference components
• Deploy fine-tuned models without code changes to existing applications
• Run agentic workflows entirely on owned SageMaker infrastructure
• Includes IAM permissions, token generation, and deployment examples

SageMaker AI now enables drop-in OpenAI-compatible inference without custom clients or code rewrites, simplifying AI application deployment on dedicated infrastructure.