What are the hardware requirements for GLM 5.2 DQ4plus Q8?

GLM 5.2 DQ4plus Q8 requires approximately 184.6 GB of VRAM and supports a context window of 1024K tokens. Quantized variants may run on less VRAM; see the Quantized Models section on this page.

Who developed GLM 5.2 DQ4plus Q8 and how large is it?

GLM 5.2 DQ4plus Q8 is developed by mlx-community, a model with 743.4b parameters. The model is published as open weights on Hugging Face and indexed on LLM Explorer with full benchmark history.

How does GLM 5.2 DQ4plus Q8 perform on standard benchmarks?

GLM 5.2 DQ4plus Q8 has the following published scores: LMSYS ELO 1319. Compare against reference models on this page or on the LLM Explorer leaderboards.

GLM 5.2 DQ4plus Q8 by mlx-community — VRAM 184.6GB, 1024K context

Name: GLM 5.2 DQ4plus Q8
Author: mlx-community

GLM 5.2 DQ4plus Q8 is an open-source language model by mlx-community. Features: 743.4b LLM, VRAM: 184.6GB, Context: 1024K, License: mit, MoE, Quantized, LLM Explorer Score: 0.45, ELO: 1319.

Arxiv:2505.02390 4-bit Base model:quantized:zai-org/g... Base model:zai-org/glm-5.2 Conversational Glm moe dsa Mlx Moe Q8 Quantized Region:us Safetensors Sharded Tensorflow

Model Card on HF 🤗: https://huggingface.co/mlx-community/GLM-5.2-DQ4plus-q8

GLM 5.2 DQ4plus Q8 Benchmarks

LMSys ELO 1319 vs 1272 (so35) 3.7%

^nn.n% — How the model compares to the reference models: Anthropic Sonnet 3.5 ("so35"), GPT-4o ("gpt4o") or GPT-4 ("gpt4").

What is the LLM Explorer Rank (Score)

GLM 5.2 DQ4plus Q8 Parameters and Internals

LLM Name	GLM 5.2 DQ4plus Q8
Repository 🤗	https://huggingface.co/mlx-community/GLM-5.2-DQ4plus-q8
Base Model(s)	zai-org/GLM-5.2 zai-org/GLM-5.2
Model Size	743.4b
Required VRAM	184.6 GB
Updated	2026-08-02
Maintainer	mlx-community
Model Type	glm_moe_dsa
Model Files	4.3 GB: 1-of-113 4.7 GB: 2-of-113 3.6 GB: 3-of-113 4.7 GB: 4-of-113 4.7 GB: 5-of-113 3.6 GB: 6-of-113 4.2 GB: 7-of-113 4.2 GB: 8-of-113 3.6 GB: 9-of-113 4.2 GB: 10-of-113 4.3 GB: 11-of-113 3.6 GB: 12-of-113 4.7 GB: 13-of-113 4.2 GB: 14-of-113 3.6 GB: 15-of-113 4.2 GB: 16-of-113 4.3 GB: 17-of-113 3.6 GB: 18-of-113 4.2 GB: 19-of-113 4.7 GB: 20-of-113 3.6 GB: 21-of-113 4.2 GB: 22-of-113 4.3 GB: 23-of-113 3.6 GB: 24-of-113 4.2 GB: 25-of-113 4.2 GB: 26-of-113 3.6 GB: 27-of-113 4.7 GB: 28-of-113 4.3 GB: 29-of-113 3.6 GB: 30-of-113 4.2 GB: 31-of-113 4.2 GB: 32-of-113 3.6 GB: 33-of-113 4.2 GB: 34-of-113 4.7 GB: 35-of-113 3.6 GB: 36-of-113 4.2 GB: 37-of-113 4.2 GB: 38-of-113 3.6 GB: 39-of-113 4.2 GB: 40-of-113 4.3 GB: 41-of-113 3.6 GB: 42-of-113 4.7 GB: 43-of-113 4.2 GB: 44-of-113 3.6 GB: 45-of-113
Quantization Type	q8
Model Architecture	GlmMoeDsaForCausalLM
License	mit
Context Length	1048576
Model Max Length	1048576
Transformers Version	5.12.0
Vocabulary Size	154880

Rank the GLM 5.2 DQ4plus Q8 capabilities

Have you tried this model? Rate its performance — this feedback helps the ML community find the right model for their needs.

Instruction Following and Task Automation
Factuality and Completeness of Knowledge
Censorship and Alignment
Data Analysis and Insight Generation
Text Generation
Text Summarization and Feature Extraction
Code Generation
Multi-Language Support and Translation

Best Alternatives to GLM 5.2 DQ4plus Q8

Best Alternatives	Context / RAM	Downloads	Likes
GLM 5.2 4bit	1024K / 207.5 GB	9403	5
GLM 5.2 Mxfp4	1024K / 233.5 GB	18445	22
GLM 5.2 MLX Nvfp4	1024K / 207.5 GB	4741	3
Aesop V1	1024K / 210.9 GB	905	4

Note: green Score (e.g. "73.2") means that the model is better than mlx-community/GLM-5.2-DQ4plus-q8.