Amazon Bedrock extended support for reinforcement fine-tuning (RFT) to open-weight models including OpenAI GPT-OSS and Qwen, with OpenAI-compatible APIs. This enables developers to customize models effectively by learning from feedback on multiple responses using only small prompt sets rather than large labeled datasets, reducing complexity and resource demands.
The fully managed process features verifiable grading systems and supports custom logic via AWS Lambda, allowing real-time performance evaluation and iteration. The capability is designed to assist organizations in adapting foundation models to specific tasks without requiring deep machine learning expertise, while maintaining data security within AWS environments.
Source & further reading
Amazon Bedrock reinforcement fine-tuning adds support for open-weight models with OpenAI-compatible APIs ↗Source published 2026-02-17. This report was prepared with AI assistance from the linked primary-source material. Claims and limitations should be checked against the original announcement or research.