【帖子标题】：zai-org/GLM-5.3 · Hugging Face
【帖子标题】：zai-org/GLM-5.3 · Hugging Face
【帖子正文】：
GLM-5.3 uses the same base model as GLM-5.2 — every gain comes from post-training. Compared with GLM-5.2, it is much better at complex coding and long-horizon tasks:
GLM-5.3采用与GLM-5.2相同的基座模型——所有性能提升均来自训练后优化。相比GLM-5.2，它在复杂编程和长周期任务上表现显著提升：
Stronger Coding: GLM-5.3 is the most capable open-weights model for coding, with a 50% improvement over GLM-5.2 on our in-house
更强编程能力：GLM-5.3是目前最具实力的开源权重编程模型，在我们内部
Z.ai
Z.ai
Code Bench. It also achieve open-source SOTA on public benchmarks including Terminal Bench 3.0 and Agents’ Last Exam.
代码基准测试中比GLM-5.2提升50%。同时在Terminal Bench 3.0和Agents’ Last Exam等公开基准测试中达到开源模型的最先进水平。
Emergent Cyber Capability: As we scaled post-training, cyber capability developed faster than we expected. GLM-5.3 is state of the art on CyberGym for vulnerability discovery, and its gains are largest further up the exploitation chain, where it more than doubles GLM-5.2 on exploitation benchmarks.
新兴网络能力：随着训练后优化的扩展，其网络安全能力发展超出预期。GLM-5.3在CyberGym漏洞发现测试中达到业界顶尖水平，在漏洞利用链的高阶环节表现尤为突出，在利用基准测试中成绩是GLM-5.2的两倍以上。
https://huggingface.co/unsloth/GLM-5.3-GGUF
https://huggingface.co/unsloth/GLM-5.3-GGUF