AWS Simplifies Advanced AI Model Customization Across Bedrock and SageMaker

Details

• AWS unveiled new AI model customization tools at re:Invent 2025, featuring Reinforcement Fine-Tuning (RFT) within Amazon Bedrock and serverless model customization in Amazon SageMaker AI.
• These enhancements allow developers to work with various leading models, including Amazon Nova, DeepSeek, GPT-OSS, Llama, and Qwen, without requiring deep machine learning knowledge or infrastructure management.
• The new RFT capability in Bedrock boosts model accuracy by up to 66% over base models and automates complex workflows, while SageMaker AI’s serverless model customization accelerates project timelines from several months to just days.
• Building on AWS’s prior announcements, these updates further democratize access to advanced AI through managed, low-code, and no-code solutions.
• By streamlining customization and removing infrastructure hurdles, AWS empowers organizations to rapidly develop and refine intelligent AI systems at scale.

Impact

These updates strengthen AWS’s place in the competitive enterprise AI market, where fast, accessible model adaptation is key. Simplified fine-tuning and multi-model support could help AWS outpace rivals that still rely on more manual, resource-intensive processes. As demand for tailored, intelligent agents grows, AWS’s new offerings will likely accelerate adoption across industries.