Z.ai
GLM-5.2
前沿
231.2K下载量3.5K点赞Jun 2026发布日期200K tokens上下文MIT许可证93 卓越质量
GLM-5.2 (753.2999877929688B parameters) requires approximately 481.6 GB of VRAM with Q4_K_M quantization. As a Mixture of Experts model with 40B active parameters, it uses less memory than its total parameter count suggests. For the best balance of quality and speed, we recommend hardware with at least 554 GB of VRAM.
快速开始
— 复制粘贴即可本地运行Copy-paste commands to run GLM-5.2 on your machine.
Run
docker run --rm -it ghcr.io/ggerganov/llama.cpp:full \
--hf-repo "zai-org/GLM-5.2" \
--hf-file "GLM-5.2-Q4_K_M.gguf" \
-c 4096 -ngl 99Quick specs
Parameters753.3B (40B active)
Architecturemoe (MoE)
Context200K tokens
Modalitytext
Min RAM293.8 GB
Rec. RAM459.5 GB (Q4_K_M)
LicenseMIT
FamilyGLM
✓ Code✓ Chat✓ Reasoning
About this model
- •Native 1M-token context for repository-scale and long-horizon agentic work.
- •Improved DSA MoE architecture (256 routed experts, 8 active per token, 1 shared) with MLA-style latent attention for reduced KV cost.
- •Flexible coding effort levels for balancing latency against solution quality.
- •Successor to GLM-5.1, tuned for sustained multi-round tool use.
相关模型
量化选项
各量化级别的 VRAM 估算
No hardware detected — fit column shows raw VRAM estimates
| Quant | Bits | VRAM | Quality | Fit |
|---|---|---|---|---|
Q2_K | 2 | 293.8 GB | Low | — |
Q3_K_S | 3 | 369.1 GB | Low | — |
NVFP4 | 4 | 421.8 GB | Medium | — |
Q4_K_M | 4 | 459.5 GB | Medium | — |
Q5_K_M | 5 | 542.4 GB | High | — |
Q6_K | 6 | 617.7 GB | High | — |
Q8_0 | 8 | 806.0 GB | Very High | — |
F16 | 16 | 1544.3 GB | Maximum | — |
硬件兼容性
全部硬件的适配估算
Computing compatibility...
内存详细分析
Reference: RTX 2060 6GB
Weights459.5 GB
KV Cache19.0 GB
Runtime2.4 GB
Headroom0.6 GB
常见问题
FAQ — GLM-5.2
另请参阅