Model reference · open weights
glm-edge is an open-weight language model from zai-org. glm-edge-4b-chat (BF16) weighs 8.7 GB; the smallest configuration that runs it is RTX 3060 12 GB.
Summary of the zai-org/glm-edge-4b-chat-gguf model card, 2026-10-01. The estimate below is for another build of the family.
What it is
| Released by | zai-org |
|---|---|
| Released | 2024-11-27 |
| VRAM | 8.7 GB for the weights |
What it runs on
| Card | Requests at once | Context max | Memory |
|---|---|---|---|
| 8K each | |||
| RTX 3060 12 GB | 2 | all 8K | 11.6 GB |
| RTX 4060 Ti 16 GB | 6 | all 8K | 15.4 GB |
| RTX 3090 24 GB | 14 | all 8K | 23.4 GB |
| RTX 4090 24 GB | 13 | all 8K | 23.4 GB |
| RTX 5090 32 GB | 21 | all 8K | 31.0 GB |
| L40S 48 GB | 34 | all 8K | 44.0 GB |
| A100 80 GB | 68 | all 8K | 78.2 GB |
| H100 80 GB | 65 | all 8K | 78.1 GB |
| RTX PRO 6000 Blackwell 96 GB | 80 | all 8K | 93.8 GB |
| DGX Spark (GB10) 128 GB unified | 94 | all 8K | 107 GB |
| H200 141 GB | 124 | all 8K | 138 GB |
| B200 180 GB | 162 | all 8K | 176 GB |
| 2× RTX 3060 12 GB tensor parallel | 13 | all 8K | 11.6 GB a card |
| 2× RTX 4060 Ti 16 GB tensor parallel | 20 | all 8K | 15.4 GB a card |
| Requests at once | 8K tokens each |
|---|---|
| 1 | 10.3 GB |
| 5 | 14.3 GB |
| 8 | 17.3 GB |
| 16 | 25.4 GB |
| 32 | 41.5 GB |
| 64 | 73.7 GB |
One card, with vLLM's small-card settings.
Builds
| Build | Params | Precision | Weights | Smallest setup |
|---|---|---|---|---|
| glm-edge-4b-chat-gguf ↗ 13 builds: Q4_0 2.5 GB … F16 8.7 GB |
— | GGUF | 2.6 GB | RTX 3060 12 GB |
| glm-edge-4b-chat (above) ↗ | 4.3B | BF16 | 8.7 GB | RTX 3060 12 GB |
From the model card
中文阅读, 点击这里
The code for adapting this model is actively being integrated into the official llama.cpp. You can test it using the
following adapted version:
git clone https://github.com/piDack/llama.cpp -b support_glm_edge_model
cmake -B build -DGGML_CUDA=ON # Or enable other acceleration hardware
cmake --build build -- -j
After installation, you can start the GLM-Edge Chat model using the following command:
llama-cli -m /model.gguf -p "\nhi\n" -ngl 999
In the command-line interface, you can interact with the model by entering your requests, and the model will provide the corresponding responses.
The usage of this model’s weights is subject to the terms outlined in the LICENSE.
Quoted from the model card on Hugging Face. The full card is behind the Hugging Face link above.
How it works