Z.ai’s GLM 5.3, a 753B-parameter mixture-of-experts model, now available on Amazon Bedrock

Z.ai’s GLM 5.3, a 753B-parameter mixture-of-experts model, is now available on Amazon Bedrock, a managed service for generative AI applications and agents. The model is optimized for coding and long-horizon agentic tasks, including multi-step reasoning, tool-augmented workflows, and sustained context across large code bases.

According to Z.ai, GLM 5.3 demonstrates competitive performance on coding benchmarks such as DeepSWE, Terminal Bench 3.0, and FrontierSWE, and reports a 50% improvement over its predecessor, GLM 5.2, on an internal coding benchmark. The model also shows emergent cybersecurity capabilities, with Z.ai measuring a leading score of 84.5 on the CyberGym benchmark at release.

Access to GLM 5.3 on Amazon Bedrock is provided through fully managed APIs, including OpenAI-compatible Responses and Chat Completions APIs, as well as Amazon Bedrock’s Invoke and Converse APIs. Users can invoke the model without managing inference infrastructure, and the service supports cross-Region inference and prompt caching to reduce latency and input costs.

The announcement highlights GLM 5.3’s integration with Amazon Bedrock’s broader feature set, including implicit (automatic) and explicit prompt caching, and improved feature parity across API endpoints. The model is available to eligible enterprise customers, positioning it for use in complex systems engineering and defensive security workflows.