Urgent.News

What's breaking now, across thousands of outlets.

AI

Z.ai’s GLM-5.3 goes open weight, but its new license aims at hyperscalers

Earlier in August, Z.ai, the Chinese AI lab behind the viral ox-alpha model that turned out to be GLM-5.3-Flash, launched The post Z.ai’s GLM-5.3 goes open weight, but its new license aims at hyperscalers appeared first on The New Stack .

Z.ai’s GLM-5.3 goes open weight, but its new license aims at hyperscalers

Z.ai, a Chinese AI lab, has released the weights of its GLM-5.3 model on Hugging Face. This move may soon place the model under the ownership of Nvidia, a major player in the tech industry. The company's new GLM-5.3 license restricts commercial use for companies with an annual revenue of over $10 billion, requiring them to pass a security review before using the model or its derivatives.

Despite this, the model maintains strong performance, ranking highly in vulnerability discovery benchmarks. Z.ai emphasizes that the model underwent extensive safety evaluations before release. The launch of GLM-5.3 has sparked discussions about the implications of restrictive licensing in the rapidly evolving landscape of open-source AI models.

Written by urgent.news from The New Stack's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.

Read the original at thenewstack.io →

More in AI

Using Scikit-Learn Pipelines: A Cleaner Way to Build Machine Learning Models

If you've spent some time building machine learning models with Python, you've probably had a notebook that looked something like this: X_train = scaler . fit_transform ( X_train ) X_test = scaler .

  • Pipelines connect multiple ML steps into a single workflow
  • Prevent data leakage by applying preprocessing only to training data
  • Keep preprocessing and modeling together for cleaner, maintainable code

Mesmo com GraphRAG, o agent se perde sem contrato de memória

Pessoal, o agent puxou um trecho bonito. Similaridade alta. Confiança alta. Estava no stub errado. Não foi “RAG fraco”. Foi porta errada. Índice caro, lixo na entrada.

  • Agent struggles without memory contract despite GraphRAG technology
  • High cost of index and garbage input identified as problems
  • Contract consists of single source, unique canonic pages, and scopes

More from Friday 28 August →