meshllm/GLM-4.7-Flash-MTP-GGUF overview

---

license: mit

library_name: mesh-llm

base_model:

• "zai-org/GLM-4.7-Flash"

pipeline_tag: text-generation

language:

• en
• zh

tags:

• gguf
• mesh-llm
• skippy
• mtp
• speculative-decoding
• distributed-inference
• local-inference
• openai-compatible

---

<div align="center">

<a href="https://www.meshllm.cloud">

<img src="https://meshllm.cloud/assets/images/jelly-logo-wordmark.png" alt="Mesh LLM" width="220">

</a>

<h1>GLM-4.7-Flash-MTP-GGUF</h1>

<p>

<strong>MTP GGUF artifact for Mesh LLM</strong>

</p>

<p>

<a href="https://www.meshllm.cloud"><img alt="Website" src="https://img.shields.io/badge/Website-meshllm.cloud-111111?style=for-the-badge"></a>

<a href="https://github.com/Mesh-LLM/mesh-llm"><img alt="GitHub" src="https://img.shields.io/badge/GitHub-Mesh--LLM-24292f?style=for-the-badge&logo=github"></a>

<a href="https://discord.gg/rs6fmc63eN"><img alt="Discord" src="https://img.shields.io/badge/Discord-Join-5865F2?style=for-the-badge&logo=discord&logoColor=white"></a>

</p>

</div>

GGUF artifact for running **GLM-4.7 Flash native multi-token prediction

(MTP)** with Mesh LLM.

This repository contains the Q4_K_M GGUF artifact for GLM-4.7 Flash with native

MTP metadata and tensors.

Highlights

| Run locally | Native MTP | OpenAI-compatible | Artifact variant |

|---|---|---|---|

| Private inference on your hardware | GLM-4.7 Flash N+1 MTP support | Serve /v1/chat/completions locally | Q4_K_M GGUF |

Model Overview

| Property | Value |

|---|---|

| Source model | zai-org/GLM-4.7-Flash |

| Model id | meshllm/GLM-4.7-Flash-MTP-GGUF |

| Family | GLM |

| Parameter scale | 31.2B reported by Hub GGUF metadata |

| Quantization | Q4_K_M |

| GGUF architecture | deepseek2 |

| Context length | 202,752 |

| Artifact size | 18.9 GB |

| Source file | GLM-4.7-Flash-MTP-Q4_K_M.gguf |

| Artifact repo | meshllm/GLM-4.7-Flash-MTP-GGUF |

Recommended Use

• Local and private inference with Mesh LLM.
• Native GLM-4.7 Flash MTP serving.
• OpenAI-compatible chat/completions workflows through Mesh LLM's local API.

For upstream architecture details, chat template guidance, sampling

recommendations, license terms, and benchmark notes, see the source model card:

zai-org/GLM-4.7-Flash.

Quickstart

# Download the Q4_K_M MTP GGUF artifact.
hf download meshllm/GLM-4.7-Flash-MTP-GGUF \
  --include 'GLM-4.7-Flash-MTP-Q4_K_M.gguf'

# Run locally with Mesh LLM.
mesh-llm serve --model "meshllm/GLM-4.7-Flash-MTP-GGUF"

# Check the local OpenAI-compatible model list.
curl -s http://localhost:3131/v1/models

# Send an OpenAI-compatible chat request.
curl -s http://localhost:3131/v1/chat/completions \
  -H "Content-Type: application/json" \
  -d '{
    "model": "meshllm/GLM-4.7-Flash-MTP-GGUF",
    "messages": [{"role": "user", "content": "Write a tiny hello-world function in Rust."}],
    "max_tokens": 128
  }'

Artifact Variant

| Property | Value |

|---|---|

| Format | gguf |

| Canonical source ref | meshllm/GLM-4.7-Flash-MTP-GGUF@main/GLM-4.7-Flash-MTP-Q4_K_M.gguf |

| Quantization | Q4_K_M |

| BOS token | [gMASK] |

| EOS token | <\|endoftext\|> |

| Conversion revision note | 60a9f590d335a032fb79e4a9fc07bce8212a33ba |

What Is Included

| Artifact | Path | Contents |

|---|---|---|

| GGUF model | GLM-4.7-Flash-MTP-Q4_K_M.gguf | Q4_K_M GLM-4.7 Flash GGUF with native MTP tensors and metadata |

| Original note | README.txt | Proof-artifact note |

| Q4 inspection | glm47-mtp-q4-gguf.txt | GGUF inspection output for the Q4_K_M artifact |

| F16 inspection | glm47-mtp-f16-gguf.txt | GGUF inspection output for the F16 conversion artifact |

| Conversion revision | llama-cpp-revision.txt | Revision recorded for the MTP GGUF conversion path |

Why MTP Matters

MTP, or multi-token prediction, lets a model predict the next token and an

extra future token in the same decode step. That gives the runtime a candidate

sequence to verify instead of waiting for one full model pass per token.

For GLM-4.7 Flash, native MTP proposes N+1: one ordinary next token plus one

additional predicted token. Mesh LLM can use that extra token as a speculative

candidate during local inference.

Validation

The repository includes GGUF inspection notes for the published artifact:

• glm47-mtp-q4-gguf.txt
• glm47-mtp-f16-gguf.txt
• llama-cpp-revision.txt

Links

• Source model: zai-org/GLM-4.7-Flash
• Mesh LLM website: meshllm.cloud
• Mesh LLM: github.com/Mesh-LLM/mesh-llm
• Discord: discord.gg/rs6fmc63eN
• Package catalog: meshllm/catalog