Z.ai reports a 753B-parameter mixture-of-experts model tuned for long-horizon agentic tasks and defensive security work.
Z.ai’s GLM 5.3 is now available on Amazon Bedrock for eligible enterprise customers. The model is a 753B-parameter mixture-of-experts system published on Hugging Face Hub and optimized for coding and long-horizon agentic tasks. Amazon Bedrock serves it through fully managed APIs with cross-Region inference, prompt caching, and service tiers, so customers run no inference infrastructure of their own.
GLM 5 arrived on Amazon Bedrock earlier this year, and GLM 5.3 builds on that lineage. Z.ai reports competitive performance on coding benchmarks including DeepSWE, Terminal Bench 3.0, and FrontierSWE, plus a 50 percent gain over GLM 5.2 on its internal coding benchmark. Z.ai also measured a leading score of 84.5 on CyberGym at release.
Teams that previously had to provision and run their own inference stack to use open-weight models can now reach GLM 5.3 through managed APIs. That shift matters for coding and agentic workloads that resend large system prompts or repository context each turn, where prompt caching cuts cost and latency. Defensive security workflows also fit, since Z.ai reports strong results on security tasks.
Builders can invoke GLM 5.3 through the OpenAI-compatible Responses and Chat Completions APIs or the Bedrock Invoke and Converse APIs. Implicit prompt caching runs by default, while explicit cache controls on the Responses and Chat Completions APIs give finer control. The AWS post pairs the model with Strix, an open-source AI penetration testing agent, for authorized security tests of an application.
What matters
- Amazon Bedrock now serves GLM 5.3 through managed APIs with cross-Region inference and prompt caching.
- Builders gain coding and agentic capacity without provisioning or operating their inference infrastructure.
- Z.ai has not published direct comparisons to GLM 5, and benchmark tests changed after GLM 5.1.
Why it matters
Z.ai has not published direct comparisons to GLM 5, and benchmark tests changed after GLM 5.1.
This GenAI News article was prepared in original wording using reporting and materials published by AWS Machine Learning Blog. Source reference: https://aws.amazon.com/blogs/machine-learning/introducing-glm-5-3-on-amazon-bedrock/.
Drafted by the GenAI News review pipeline.