Announcing Amazon SageMaker Inference for custom Amazon Nova models

Amazon SageMaker Inference now supports GA deployment of custom Amazon Nova models for production-grade inference.

• •Covers Nova Micro, Nova Lite, and Nova 2 Lite models with continued pre-training, SFT, or reinforcement fine-tuning
• •Uses EC2 G5/G6 instances (more cost-efficient than P5) with auto-scaling based on 5-minute usage patterns
• •Configurable parameters include context length, concurrency, batch size, temperature, top_p, and reasoning effort
• •Deploy via SageMaker Studio UI or SDK using create_model, create_endpoint_config, and create_endpoint APIs
• •Available in us-east-1 and us-west-2 with per-hour billing and no minimum commitment