Build faster, more cost-efficient, highly accurate models with Amazon Bedrock Model Distillation (preview)

AWS has announced Amazon Bedrock Model Distillation, a new preview feature that automates creating smaller, faster, and more cost-efficient AI models while maintaining high accuracy.

• Enables creating distilled models up to 5x faster and 75% less expensive than original large models
• Uses knowledge transfer from a "teacher" model to a smaller "student" model
• Supports models from Anthropic, Meta, and Amazon
• Generates synthetic data to improve model performance
• Available in US East (N. Virginia) and US West (Oregon) regions

The feature helps customers create more efficient generative AI models for specific use cases with minimal accuracy loss, addressing challenges of model size, latency, and cost.