README.md

---
language:
  - en
  - zh
library_name: transformers
license: mit
pipeline_tag: text-generation
---

# GLM-5-FP8

<div align="center">
<img src=https://raw.githubusercontent.com/zai-org/GLM-5/refs/heads/main/resources/logo.svg width="15%"/>
</div>
<p align="center">
    👋 Join our <a href="https://raw.githubusercontent.com/zai-org/GLM-5/refs/heads/main/resources/wechat.png" target="_blank">WeChat</a> or <a href="https://discord.gg/QR7SARHRxK" target="_blank">Discord</a> community.
    <br>
    📖 Check out the GLM-5 <a href="https://z.ai/blog/glm-5" target="_blank">technical blog</a>.
    <br>
    📍 Use GLM-5 API services on <a href="https://docs.z.ai/guides/llm/glm-5">Z.ai API Platform. </a>
    <br>
    👉 One click to <a href="https://chat.z.ai">GLM-5</a>.
</p>

## Introduction

We are launching GLM-5, targeting complex systems engineering and long-horizon agentic tasks. Scaling is still one of the most important ways to improve the intelligence efficiency of Artificial General Intelligence (AGI). Compared to GLM-4.5, GLM-5 scales from 355B parameters (32B active) to 744B parameters (40B active), and increases pre-training data from 23T to 28.5T tokens. GLM-5 also integrates DeepSeek Sparse Attention (DSA), largely reducing deployment cost while preserving long-context capacity.

Reinforcement learning aims to bridge the gap between competence and excellence in pre-trained models. However, deploying it at scale for LLMs is a challenge due to the RL training inefficiency. To this end, we developed [slime](https://github.com/THUDM/slime), a novel **asynchronous RL infrastructure** that substantially improves training throughput and efficiency, enabling more fine-grained post-training iterations. With advances in both pre-training and post-training, GLM-5 delivers significant improvement compared to GLM-4.7 across a wide range of academic benchmarks and achieves best-in-class performance among all open-source models in the world on reasoning, coding, and agentic tasks,  closing the gap with frontier models.

## Benchmark

|                                  | GLM-5                  | GLM-4.7   | DeepSeek-V3.2 | Kimi K2.5 | Claude Opus 4.5 | Gemini 3 Pro | GPT-5.2 (xhigh) |
| -------------------------------- | ---------------------- | --------- | ------------- |-----------| --------------- | ------------ | --------------- |
| HLE                              | 30.5                   | 24.8      | 25.1          | 31.5      | 28.4            | 37.2         | 35.4            |
| HLE (w/ Tools)                   | 50.4                   | 42.8      | 40.8          | 51.8      | 43.4*           | 45.8*        | 45.5*           |
| AIME 2026 I                      | 92.7                   | 92.9      | 92.7          | 92.5      | 93.3            | 90.6         | -               |
| HMMT Nov. 2025                   | 96.9                   | 93.5      | 90.2          | 91.1      | 91.7            | 93.0         | 97.1            |
| IMOAnswerBench                   | 82.5                   | 82.0      | 78.3          | 81.8      | 78.5            | 83.3         | 86.3            |
| GPQA-Diamond                     | 86.0                   | 85.7      | 82.4          | 87.6      | 87.0            | 91.9         | 92.4            |
| SWE-bench Verified               | 77.8                   | 73.8      | 73.1          | 76.8      | 80.9            | 76.2         | 80.0            |
| SWE-bench Multilingual           | 73.3                   | 66.7      | 70.2          | 73.0      | 77.5            | 65.0         | 72.0            |
| Terminal-Bench 2.0 (Terminus 2)  | 56.2 / 60.7 † | 41.0      | 39.3          | 50.8      | 59.3            | 54.2         | 54.0            |
| Terminal-Bench 2.0 (Claude Code) | 56.2 / 61.1 †  | 32.8      | 46.4          | -         | 57.9            | -            | -               |
| CyberGym                         | 43.2                   | 23.5      | 17.3          | 41.3      | 50.6            | 39.9         | -               |
| BrowseComp                       | 62.0                   | 52.0      | 51.4          | 60.6      | 37.0            | 37.8         | -               |
| BrowseComp (w/ Context Manage)   | 75.9                   | 67.5      | 67.6          | 74.9      | 67.8            | 59.2         | 65.8            |
| BrowseComp-Zh                    | 72.7                   | 66.6      | 65.0          | 62.3      | 62.4            | 66.8         | 76.1            |
| τ²-Bench                         | 89.7                   | 87.4      | 85.3          | 80.2      | 91.6            | 90.7         | 85.5            |
| MCP-Atlas (Public Set)           | 67.8                   | 52.0      | 62.2          | 63.8      | 65.2            | 66.6         | 68.0            |
| Tool-Decathlon                   | 38.0                   | 23.8      | 35.2          | 27.8      | 43.5            | 36.4         | 46.3            |
| Vending Bench 2                  | $4,432.12              | $2,376.82 | $1,034.00     | $1,198.46 | $4,967.06       | $5,478.16    | $3,591.33       |

> *: refers to their scores of full set.
> 
> †: A verified version of Terminal-Bench 2.0 that fixes some ambiguous instructions.
See footnote for more evaluation details.

### Footnote

* **Humanity’s Last Exam (HLE) & other reasoning tasks**: We evaluate with a maximum generation length of 131,072 tokens (`temperature=1.0, top_p=0.95, max_new_tokens=131072`). By default, we report the text-only subset; results marked with * are from the full set. We use GPT-5.2 (medium) as the judge model. For HLE-with-tools, we use a maximum context length of 202,752 tokens.
* **SWE-bench & SWE-bench Multilingual**: We run the SWE-bench suite with OpenHands using a tailored instruction prompt. Settings: `temperature=0.7, top_p=0.95, max_new_tokens=16384`, with a 200K context window.
* **BrowserComp**: Without context management, we retain details from the most recent 5 turns. With context management, we use the same discard-all strategy as DeepSeek-v3.2 and Kimi K2.5.
* **Terminal-Bench 2.0 (Terminus 2)**: We evaluate with the Terminus framework using `timeout=2h, temperature=0.7, top_p=1.0, max_new_tokens=8192`, with a 128K context window. Resource limits are capped at 16 CPUs and 32 GB RAM.
* **Terminal-Bench 2.0 (Claude Code)**: We evaluate in Claude Code 2.1.14 (think mode, default effort) with `temperature=1.0, top_p=0.95, max_new_tokens=65536`. We remove wall-clock time limits due to generation speed, while preserving per-task CPU and memory constraints. Scores are averaged over 5 runs. We fix environment issues introduced by Claude Code and also report results on a verified Terminal-Bench 2.0 dataset that resolves ambiguous instructions (see: [https://huggingface.co/datasets/zai-org/terminal-bench-2-verified](https://huggingface.co/datasets/zai-org/terminal-bench-2-verified)).
* **CyberGym**: We evaluate in Claude Code 2.1.18 (think mode, no web tools) with (`temperature=1.0, top_p=1.0, max_new_tokens=32000`) and a 250-minute timeout per task. Results are single-run Pass@1 over 1,507 tasks.
* **MCP-Atlas**: All models are evaluated in think mode on the 500-task public subset with a 10-minute timeout per task. We use Gemini 3 Pro as the judge model.
* **τ²-bench**: We add a small prompt adjustment in Retail and Telecom to avoid failures caused by premature user termination. For Airline, we apply the domain fixes proposed in the Claude Opus 4.5 system card.
* **Vending Bench 2**: Runs are conducted independently by [Andon Labs](https://andonlabs.com/evals/vending-bench-2).


## Serve GLM-5 Locally

### Prepare environment

vLLM, SGLang, and xLLM all support local deployment of GLM-5. A simple deployment guide is provided here.

+ vLLM

    Using Docker as:

    ```shell
    docker pull vllm/vllm-openai:nightly 
    ```

    or using pip:

    ```shell
    pip install -U vllm --pre --index-url https://pypi.org/simple --extra-index-url https://wheels.vllm.ai/nightly
    ```

    then upgrade transformers:

    ```
    pip install git+https://github.com/huggingface/transformers.git
    ```

+ SGLang

    Using Docker as:
    ```bash
    docker pull lmsysorg/sglang:glm5-hopper # For Hopper GPU
    docker pull lmsysorg/sglang:glm5-blackwell # For Blackwell GPU
    ```

### Deploy

+ vLLM

    ```shell
    vllm serve zai-org/GLM-5-FP8 \
         --tensor-parallel-size 8 \
         --gpu-memory-utilization 0.85 \
         --speculative-config.method mtp \
         --speculative-config.num_speculative_tokens 1 \
         --tool-call-parser glm47 \
         --reasoning-parser glm45 \
         --enable-auto-tool-choice \
         --served-model-name glm-5-fp8
    ```

    Check the [recipes](https://github.com/vllm-project/recipes/blob/main/GLM/GLM5.md) for more details.

+ SGLang

    ```shell
    python3 -m sglang.launch_server \
      --model-path zai-org/GLM-5-FP8 \
      --tp-size 8 \
      --tool-call-parser glm47  \
      --reasoning-parser glm45 \
      --speculative-algorithm EAGLE \
      --speculative-num-steps 3 \
      --speculative-eagle-topk 1 \
      --speculative-num-draft-tokens 4 \
      --mem-fraction-static 0.85 \
      --served-model-name glm-5-fp8
    ```
    
    Check the [sglang cookbook](https://cookbook.sglang.io/autoregressive/GLM/GLM-5) for more details.

+ xLLM and other Ascend NPU

    Please check the deployment guide [here](https://github.com/zai-org/GLM-5/blob/main/example/ascend.md).


## Citation

Our technical report is coming soon.