Amazon Bedrock Model Evaluation LLM-as-a-judge is now generally available

Amazon Bedrock Model Evaluation's LLM-as-a-judge capability is now generally available, offering comprehensive model evaluation tools.

• Allows evaluating and comparing models using other LLMs as judges
• Supports quality metrics like correctness, completeness, and professional tone
• Includes responsible AI metrics such as harmfulness and answer refusal
• Can evaluate all Amazon Bedrock models, including serverless and marketplace models
• New feature allows "bring your own inference responses" from any model or system

The service provides human-like evaluation quality at a lower cost and significantly reduces evaluation time, making model selection more efficient and comprehensive.