公告
先看证据,再决定买不买

频道每天最多 3 条价格异动与中转状态;具体商品请用机器人设置降价/补货提醒。交流群提问请带预算、模型、工具和使用频率。

查看
社群与联系Telegram 群点击加入Telegram 频道每天最多 3 条有效价格情报联系我们tgAIPricedb交流群979789483
返回资讯列表
product

智谱 GLM-5.3 开放权重发布,可下载、运行和定制

Z.ai 宣布 GLM-5.3 开放权重,定位于智能体编程和网络防御任务。模型权重已在 Hugging Face 提供,并配套发布技术博客。

94% VERIFIED

Z.ai 宣布推出 GLM-5.3,并将其作为开放权重模型提供。用户可以下载模型权重,在本地或其他受支持的环境中运行,并根据需求进行定制。

官方将 GLM-5.3 定位为面向智能体编程和网络防御的高能力模型。相关权重托管在 Hugging Face,Z.ai 同时发布了技术博客,介绍模型及其使用信息。

目前公开材料主要能够确认模型发布和获取渠道;有关具体性能、基准测试及授权条款的细节,仍应以模型页面和官方技术文档为准。

来源证据

GLM-5.3 is now open-weight. Our most capable model for...daily.dev · supporting

GLM-5.3 is now open-weight. Our most capable model for agentic coding and cyber defense is now available to download, run, and customize.

John Paul Fusin - GLM-5.3 is now available as an...facebook.com · supporting

## John Paul Fusin's post ### John Paul Fusin 6d · GLM-5.3 is now available as an open-weight model for coding, agentic workflows, and cyber defense. Its weights can be downloaded, run, and customized, with local serving paths listed for SGLang, vLLM, Transformers, and other frameworks. Key features: • Post-training focused on complex coding and long-horizon tasks. • Z ai reports a 50% improvement over GLM-5.2 on its in-house Code Bench. • Official benchmark tables cover Terminal Bench 3.0, DeepSWE, CyberGym, and agentic-task evaluations. • The Hugging Face model card lists English and Chinese support, plus Safetensors and FP8 files. May be a graphic of text that says

Z.ai's Postlinkedin.com · supporting

GLM-5.3 is now open-weight. Our most capable model for agentic coding and cyber defense is now available to download, run, and customize.

Z.ai "GLM-5.3 is now open-weight. Our most capable ...reddit.com · supporting

Z.ai "GLM-5.3 is now open-weight. Our most capable model for agentic coding and cyber defense is now available to download, run, and customize.

Z.ai on X: "GLM-5.3 is now open-weight. Our most capable model for agentic coding and cyber defense is now available to download, run, and customize. Weights: https://t.co/v1IbWMXxg4 Tech blog: https://t.co/ekQkO83jCv" / Xx.com · supporting

Log inSign up ## Post # Z.ai on X: "GLM-5.3 is now open-weight. Our most capable model for agentic coding and cyber defense is now available to download, run, and customize. Weights: Tech blog: @Zai_org Z.ai @Zai\_org GLM-5.3 is now open-weight. Our most capable model for agentic coding and cyber defense is now available to download, run, and customize. Weights: huggingface.co/zai-org/GLM-5.3 Tech blog: z.ai/blog/glm-5.3 3:04 PM · Aug 28, 20261.4MViews 274 @Zai_org Z.ai @Zai\_org GLM-5.3 is now open-weight. Our most capable model for agentic coding and cyber defense is now available to download, run, and customize. Weights: huggingface.co/zai-org/GLM-5.3 Tech blog: z.ai/blog/glm-5.3 3:04 PM · Aug 28, 20261.4MViews 274 @UnslothAI Unsloth AI @UnslothAI [...] 274 @UnslothAI Unsloth AI @UnslothAI Aug 28 Congrats Z ai team, once again another amazing open-source release! 🥰 We're working on GLM-5.3 Unsloth GGUFs for those who can run it l

Z.ai releases GLM-5.3 as open-weight · 268 HN comments | Zelizeli.app · supporting

# Z.ai releases GLM-5.3 as open-weight model GLM-5.3 is now open-weight ▲ 780jeudesprits268 HN discussion Z.ai releases GLM-5.3 as open-weight model Z.ai has released GLM-5.3, its most capable model for agentic coding and cyber defense, now available as open weights for download, running, and customization. The model is hosted on Hugging Face, with a tech blog detailing its capabilities. The release has sparked community responses, including Unsloth AI working on GGUF versions for local execution and Baseten planning to host it. Notably, GLM-5.3 has moved away from the MIT license to a semi-non-commercial license, similar to other Chinese AI models. [...] > GLM-5.3 is now open-weight. Our most capable model for agentic coding and cyber defense is now available to download, run, and customize. Read original →HN discussionRSS ## HN discussion 268 1. revolvingthrow GLM 5.3 is probably the sweet spot open weights model if you want to go beyond deepseek flash or the new glm flas