---
license: other
license_name: glm-4
license_link: https://huggingface.co/THUDM/glm-4-9b-chat/blob/main/LICENSE
inference: false
base_model: THUDM/glm-4-9b-chat
model_creator: THUDM
model_name: glm-4-9b-chat
quantized_by: Second State Inc.
language:
- zh
- en
tags:
- glm
- chatglm
- thudm
---

<!-- header start -->
<!-- 200823 -->
<div style="width: auto; margin-left: auto; margin-right: auto">
<img src="https://github.com/LlamaEdge/LlamaEdge/raw/dev/assets/logo.svg" style="width: 100%; min-width: 400px; display: block; margin: auto;">
</div>
<hr style="margin-top: 1.0em; margin-bottom: 1.0em;">
<!-- header end -->

# glm-4-9b-chat-GGUF

## Original Model

[THUDM/glm-4-9b-chat](https://huggingface.co/THUDM/glm-4-9b-chat)

## Run with LlamaEdge

- LlamaEdge version: [v0.12.3](https://github.com/LlamaEdge/LlamaEdge/releases/tag/0.12.3) and above

- Prompt template

  - Prompt type: `glm-4-chat`

  - Prompt string

    ```text
    [gMASK]<|system|>
    {system_message}<|user|>
    {user_message_1}<|assistant|>
    {assistant_message_1}
    ```

- Context size: `128000`

- Run as LlamaEdge service

  ```bash
  wasmedge --dir .:. --nn-preload default:GGML:AUTO:glm-4-9b-chat-Q5_K_M.gguf \
    llama-api-server.wasm \
    --prompt-template glm-4-chat \
    --ctx-size 128000 \
    --model-name glm-4-9b-chat
  ```

- Run as LlamaEdge command app

  ```bash
  wasmedge --dir .:. \
    --nn-preload default:GGML:AUTO:glm-4-9b-chat-Q5_K_M.gguf \
    llama-chat.wasm \
    --prompt-template glm-4-chat \
    --ctx-size 128000
  ```

## Quantized GGUF Models

| Name | Quant method | Bits | Size | Use case |
| ---- | ---- | ---- | ---- | ----- |
| [glm-4-9b-chat-Q2_K.gguf](https://huggingface.co/second-state/glm-4-9b-chat-GGUF/blob/main/glm-4-9b-chat-Q2_K.gguf)     | Q2_K   | 2 | 3.99 GB| smallest, significant quality loss - not recommended for most purposes |
| [glm-4-9b-chat-Q3_K_L.gguf](https://huggingface.co/second-state/glm-4-9b-chat-GGUF/blob/main/glm-4-9b-chat-Q3_K_L.gguf) | Q3_K_L | 3 | 5.28 GB| small, substantial quality loss |
| [glm-4-9b-chat-Q3_K_M.gguf](https://huggingface.co/second-state/glm-4-9b-chat-GGUF/blob/main/glm-4-9b-chat-Q3_K_M.gguf) | Q3_K_M | 3 | 5.06 GB| very small, high quality loss |
| [glm-4-9b-chat-Q3_K_S.gguf](https://huggingface.co/second-state/glm-4-9b-chat-GGUF/blob/main/glm-4-9b-chat-Q3_K_S.gguf) | Q3_K_S | 3 | 4.59 GB| very small, high quality loss |
| [glm-4-9b-chat-Q4_0.gguf](https://huggingface.co/second-state/glm-4-9b-chat-GGUF/blob/main/glm-4-9b-chat-Q4_0.gguf)     | Q4_0   | 4 | 5.46 GB| legacy; small, very high quality loss - prefer using Q3_K_M |
| [glm-4-9b-chat-Q4_K_M.gguf](https://huggingface.co/second-state/glm-4-9b-chat-GGUF/blob/main/glm-4-9b-chat-Q4_K_M.gguf) | Q4_K_M | 4 | 6.25 GB| medium, balanced quality - recommended |
| [glm-4-9b-chat-Q4_K_S.gguf](https://huggingface.co/second-state/glm-4-9b-chat-GGUF/blob/main/glm-4-9b-chat-Q4_K_S.gguf) | Q4_K_S | 4 | 5.75 GB| small, greater quality loss |
| [glm-4-9b-chat-Q5_0.gguf](https://huggingface.co/second-state/glm-4-9b-chat-GGUF/blob/main/glm-4-9b-chat-Q5_0.gguf)     | Q5_0   | 5 | 6.55 GB| legacy; medium, balanced quality - prefer using Q4_K_M |
| [glm-4-9b-chat-Q5_K_M.gguf](https://huggingface.co/second-state/glm-4-9b-chat-GGUF/blob/main/glm-4-9b-chat-Q5_K_M.gguf) | Q5_K_M | 5 | 7.14 GB| large, very low quality loss - recommended |
| [glm-4-9b-chat-Q5_K_S.gguf](https://huggingface.co/second-state/glm-4-9b-chat-GGUF/blob/main/glm-4-9b-chat-Q5_K_S.gguf) | Q5_K_S | 5 | 6.69 GB| large, low quality loss - recommended |
| [glm-4-9b-chat-Q6_K.gguf](https://huggingface.co/second-state/glm-4-9b-chat-GGUF/blob/main/glm-4-9b-chat-Q6_K.gguf)     | Q6_K   | 6 | 8.26 GB| very large, extremely low quality loss |
| [glm-4-9b-chat-Q8_0.gguf](https://huggingface.co/second-state/glm-4-9b-chat-GGUF/blob/main/glm-4-9b-chat-Q8_0.gguf)     | Q8_0   | 8 | 9.99 GB| very large, extremely low quality loss - not recommended |
| [glm-4-9b-chat-f16.gguf](https://huggingface.co/second-state/glm-4-9b-chat-GGUF/blob/main/glm-4-9b-chat-f16.gguf)       | f16    | 16 | 18.8 GB|  |

*Quantized with llama.cpp b3333*