Z.ai has formally launched GLM-5.3-Flash, revealing that the previously previewed "Ox Alpha" model is its public identity. The model features 320 billion total parameters with 18 billion active, a 1 million-token context window, and native multimodal capabilities.

  • Released under the MIT License, it is available via weights on Hugging Face, API, chat, and coding plans.
  • Z.ai claims it outperforms GLM-5.2 at every effort level and matches Claude Opus 4.8 on coding benchmarks.
  • Artificial Analysis reports an Intelligence Index score of 57, tying GPT-5.6 Terra and Muse Spark 1.2 but at significantly lower cost.
  • The architecture utilizes Kimi Linear-style hybrid attention, DeepSeek V4-style residual paths, and runs entirely on Chinese AI chips.
  • Early infrastructure support appeared from CoreWeave and Baseten, with a prompt template correction issued shortly after launch.

The release positions GLM-5.3-Flash as a highly price-competitive option for code generation and agentic tasks, though independent evaluations suggest it may be less accurate on broad factual knowledge compared to its reasoning performance.