1.5TB of model, 350GB of computer. Math shouldn't work like this.

Researchers reveal that Chinese AI model GLM-5.3 has successfully executed complex cyberattacks in controlled tests, ...

I have summarized the overview of "Project Maya".1. IntroductionIn local LLMs, GPU VRAM capacity is a major constraint.

GLM-4.5 represents a technological breakthrough as Z.ai's first attempts into the open-source community with a MoE architecture, available in two powerful models. The flagship GLM-4.5 boasts 355 ...

Irregular tests GLM-5.3’s cyber capabilities, with multistage attack successes and instruction violations in separate ...

Anthropic's red team found Z.ai's GLM-5.3 could exploit six Firefox versions using unshipped fixes, and Mozilla patched a ...

BEIJING, April 15, 2025 /PRNewswire/ -- Z.ai (formerly Zhipu) announces the open-sourcing of its 32B and 9B GLM model series, including base, reasoning, and rumination models, all under the MIT ...

Z.AI's open-weight GLM-5.3 completed 3 of 13 multistage cyberattack scenarios in Irregular's tests and ignored instructions ...

Z.ai’s launch chart shows GLM-Legal ahead of Kimi K3 (12.92%), Claude Fable 5 (11.25%) and GPT-6 Astra (5.42%). Those figures match Vals AI. The chart omits 6 models ranked higher, led by Muse Spark 1 ...

OpenAI has done something different recently: offered banked resets that suggest it wants you to save your token splurge on a ...

The GLM-5.3 model successfully completed 3 out of 13 tested cyberattack scenarios at least once. However, this Chinese model ...