Chinese startup Zhipu AI launched GLM-5.3, which the company says is the most powerful among open-weights models for coding. The new version retains the same base as GLM-5.2, with all gains coming from extended post-training.
The company states that the biggest advances are in agent-based tasks. The model was also trained with data and environments developed to locate software vulnerabilities.
Vulnerability detection
According to Z.ai, the model "began reasoning through multiple exploitation stages, forming coherent plans for complete exploitation chains." In partnership with security teams in China, Zhipu says it has found 2,436 vulnerabilities in 269 projects, some up to 40 years old. The flaws are documented in a public registry.
GLM-5.3 is now available through the GLM Coding Plan and works with coding agents such as ZCode, Claude Code, and OpenCode. The model weights are expected to be released as open source in two weeks, after security reviews are completed.
