Zhipu AI (Z.ai)9BQ4_K_M 下约 6 GB 显存

GLM-4 9B — 显存、速度与本地部署

作者: Jakub Rusinowski · 最后更新:

GLM-4-9B, Z.ai's compact bilingual model. Strong Chinese-English performance and general reasoning; runs comfortably on 8 GB VRAM. Previously listed as 'GLM-4.7 9B', which does not exist - GLM-4.7 is a ~400B MoE whose small sibling is GLM-4.7-Flash at 30B-A3B. The Ollama tag recorded here, glm4:9b, pulls GLM-4-9B, so the record was correct apart from its name.

GLM-4 9B 在 Q4_K_M 下约需 6 GB 显存——量化权重加框架开销,不含 KV 缓存。在 Apple Silicon 上,这部分来自统一内存。

逻辑87
创意84
编程82

按量化级别的显存与速度

计算基准:NVIDIA RTX 4090 (24 GB)。仅含权重与开销:该模型架构未公开,因此未计入 KV 缓存。

量化显存速度(估算)适配
Q2_K
2.63 bpw
3.8 GB~144 tok/s可运行
Q3_K_M
3.41 bpw
4.6 GB~123 tok/s可运行
Q4_K_M
4.83 bpw
6.2 GB~98 tok/s可运行
Q5_K_M
5.67 bpw
7.2 GB~87 tok/s可运行
Q6_K
6.56 bpw
8.2 GB~78 tok/s可运行
Q8_0
8.50 bpw
10.4 GB~64 tok/s可运行
F16
16.00 bpw
18.8 GB~37 tok/s可运行

黑色标记 = NVIDIA RTX 4090 (24 GB) 上的可用显存。 估算来自内存带宽屋顶线模型,详见 方法说明页. GLM-4 9B 显存计算器 →

运行 GLM-4 9B

目录中能运行 GLM-4 9B 的最便宜 GPU 是 Intel Arc B570 (10 GB).

购买此硬件 Intel Arc B570 10GB — 10 GB VRAM · 150 W board power立即云端部署 RunPod 上的 RTX 4090 — 低至 $0.34/小时 · 价格核实于 2026-07

或在 Vast.ai 比较,低至 $0.35/小时 (typical low · varies)

作为亚马逊联盟成员,我们从符合条件的购买中获得收入。云 GPU 链接为推荐链接——我们可能获得佣金,您无需额外付费。

联盟营销声明: 本页部分链接为联盟推广链接——如果你通过它们购买,LLM Configurator 可能会获得佣金,而你无需支付任何额外费用。作为亚马逊联盟成员(Amazon Associate),LLM Configurator 会从符合条件的购买中获得收益。
Intel Arc B570 10GB
10 GB VRAM · 150 W board power
2026年价格波动较大——请以当前商品页价格为准。

如何运行 GLM-4 9B

安装 Ollama,然后运行:

ollama run glm4:9b
Hugging Face 上的权重: THUDM/glm-4-9b-chat ↗

规格

Corroborated — Two or more independent sources agree on these figures, but the model card itself was not retrieved. Treat the numbers as good rather than confirmed.

参数量
9 Billion
上下文窗口
128,000
架构
Dense
提供商
Zhipu AI (Z.ai)
许可证
Apache 2.0
规格量化
Q4_K_M
系统内存
16 GB
记录更新于
2026-02-14
许可证Apache-2.0允许商业使用

Commercial use permitted. No usage restrictions beyond attribution.

质量与使用场景

评分由模型作者或独立评测方发布——衡量质量而非吞吐量,并非我们实测。

最适合chatmultilingualreasoningchinese

我的 GPU 能运行 GLM-4 9B 吗?

GLM-4.7 / GLM-Z1 的其他尺寸

GLM-4 9B — 常见问题

How much VRAM does GLM-4 9B need?

About 6 GB at Q4_K_M — quantized weights plus framework overhead, before any KV cache. The cache grows with context length and is added on top; the table above folds it in. Apple Silicon counts unified memory toward the same figure.

Does GLM-4 9B run on an RTX 4090 (24 GB)?

Yes. GLM-4 9B needs about 6 GB at Q4_K_M, inside a 24 GB card, at an estimated 98 tokens/sec.

How do I run GLM-4 9B locally?

Install Ollama and run `ollama run glm4:9b`. That pulls the weights and starts a local OpenAI-compatible endpoint; after the download nothing leaves the machine.

What other sizes does GLM-4.7 / GLM-Z1 come in?

GLM-4 9B (6 GB), GLM-Z1 32B (Reasoning) (20 GB), GLM-4.7-Flash 30B-A3B (19 GB). Every size shares the family's training and licence; the larger ones score higher and need proportionally more memory.