Introducing GLM 5.3 on Amazon Bedrock
aws ·
Z.ai has released GLM 5.3, a 753B-parameter mixture-of-experts model, on Amazon Bedrock to support complex coding and long-horizon agentic workloads. The model features enhanced performance on coding benchmarks and notable capabilities in cybersecurity, including a leading score on the CyberGym benchmark. Users can access GLM 5.3 via fully managed APIs that include cross-Region inference, service tiers for cost or latency optimization, and prompt caching to reduce input costs. The announcement provides technical guidance for invoking the model using OpenAI-compatible APIs and Amazon Bedrock’s native interfaces. Additionally, the article demonstrates an agentic security testing workflow using Strix, an open-source penetration testing agent configured to run against local applications. This integration allows developers to perform authorized security assessments while keeping inference within their AWS account controls. Access to the model is currently available to eligible enterprise customers through the Amazon Bedrock console or programmatically.
aws