GLM-5.3

Z.ai's text-only 753B coding model: 40B active, 1M context, open weights under a license that only limits huge hosting businesses.

In short. GLM-5.3 is a large AI model from Z.ai, built mainly for writing and fixing software and for security work. It reads and writes text only. The license is free for nearly everyone, including companies, and it can be run on a big GPU server or a very large-memory machine.

Summary

GLM-5.3 is the flagship of Z.ai's GLM-5 series. Z.ai launched it on its API in August 2026 and said it would release the weights after a safety review; the weights appeared on Hugging Face on 28 August 2026. It focuses on coding and security tasks, and Z.ai reports a 50 percent gain over GLM-5.2 on its own Code Bench and a state-of-the-art result among open models on Terminal Bench 3.0 (self-reported).

It is a text-only mixture-of-experts model with about 753 billion parameters, of which roughly 40 billion are active per token. The 753 billion figure includes a built-in draft layer for speculative decoding; the main stack is about 744 billion. It has 78 layers, 256 routed experts with 8 active plus 1 shared expert, and a sparse attention indexer that reads only the 2,048 most relevant cached tokens at each step. The context window is 1,048,576 tokens.

The FP8 checkpoint is about 756 GB. BF16, NVFP4 and GGUF builds also exist. The GLM-5.3 License allows use, modification, distribution and sale. Its one condition is that model-as-a-service businesses with more than US$10 billion of revenue in any twelve months must pass a Z.ai security review before commercial use.

Good for

  • Self-hosted coding agents and large-repository work
  • Security research and code review, which Z.ai highlights
  • Organizations that want an open model with a very high revenue threshold for restrictions
  • Long-context tasks up to 1 million tokens

Watch out for

  • It is text-only; use GLM-5.3-Flash if you need image input
  • Even compressed it needs a GPU node or a machine with roughly 512 GB of fast memory
  • The license is custom, not MIT; only model-hosting businesses above US$10 billion of revenue face the extra review
  • Benchmark results and the vulnerability counts come from Z.ai itself

Performance claims come from the developer and have not been independently verified.

How much memory does it need?

Estimates for the GLM-5.3 variant. A mixture-of-experts model must hold all of its experts in memory even though only some are used for each token, so total parameters decide the memory and active parameters decide the speed.

Variants Parameters BF16 FP8 / INT8 4-bit (Q4) Fits on
GLM-5.3 FP8 753B 40B active 1.73 TB 866 GB 520 GB GPU server

Estimates: weights plus about 15% overhead. Real usage grows with context length and depends on the software.

VRAM calculator

The license in plain words

Close to MIT. The only extra condition applies to giant hosting businesses: a model-as-a-service operator with more than US$10 billion of revenue must pass a Z.AI security review. Everyone else can treat it like MIT.

  • Security review for very large hosts

More about GLM-5.3 License

See also

Frequently asked questions

What is GLM-5.3?
GLM-5.3 is a large AI model from Z.ai, built mainly for writing and fixing software and for security work. It reads and writes text only. The license is free for nearly everyone, including companies, and it can be run on a big GPU server or a very large-memory machine.
Can I use GLM-5.3 commercially?
Yes, with conditions. GLM-5.3 is released under the GLM-5.3 License, which allows commercial use but adds extra rules: Security review for very large hosts. Read the license text before you ship a product.
Is GLM-5.3 open source?
GLM-5.3 is open-weight, not open source in the strict sense. You can download the model files, but the GLM-5.3 License adds conditions that the Open Source Initiative definition does not allow.
Can I run GLM-5.3 on my own computer?
Not on a desktop. The smallest variant, GLM-5.3, needs about 520 GB at 4-bit precision, which means a GPU server with several data-center cards.
When was GLM-5.3 released?
GLM-5.3 was released by Z.ai (Zhipu AI) on August 28, 2026.
How long a context window does GLM-5.3 have?
GLM-5.3 supports a context window of up to 1M tokens.

Official links

Sources

Specifications come from these primary sources. Benchmark claims, where mentioned, are reported by the developer.

Release date and facts last verified November 8, 2026.

Related models

Model Released Parameters Context License Smallest variant needs
GLM-5.3-Flash Z.ai (Zhipu AI) 320B 18B active 1M MIT (Permissive) 221 GB Big-memory desktop
DeepSeek V4.1 Flash DeepSeek 552B 16B active 1M MIT (Permissive) 561 GB GPU server
Qwen3.8-Max Alibaba Qwen 2.45T 95B active 1M Qwen3.8-Max License (Conditional) 1.69 TB Multi-node cluster
Hydrogen-1.0 Deep Principle 753B 40B active 1M MIT (Permissive) 465 GB Big-memory desktop