What are the hardware requirements for SuperGLM 5.2 Abliterated NVFP4?

SuperGLM 5.2 Abliterated NVFP4 requires approximately 410 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 SuperGLM 5.2 Abliterated NVFP4 and how large is it?

SuperGLM 5.2 Abliterated NVFP4 is developed by Jiunsong, a model with 381b parameters. The model is published as open weights on Hugging Face and indexed on LLM Explorer with full benchmark history.

Where can I download or evaluate SuperGLM 5.2 Abliterated NVFP4?

SuperGLM 5.2 Abliterated NVFP4 is hosted on Hugging Face and linked from this page. LLM Explorer also lists quantized variants and similar alternatives if available.

SuperGLM 5.2 Abliterated NVFP4 by Jiunsong — VRAM 410GB, 1024K context

Name: SuperGLM 5.2 Abliterated NVFP4
Author: Jiunsong

SuperGLM 5.2 Abliterated NVFP4 is an open-source language model by Jiunsong. Features: 381b LLM, VRAM: 410GB, Context: 1024K, License: mit, MoE, LLM Explorer Score: 0.33.

8-bit Abliterated Base model:finetune:nvidia/glm... Base model:nvidia/glm-5.2-nvfp... Conversational Endpoints compatible Glm Glm-5.2 Glm moe dsa Modelopt Moe Nvfp4 Obliteratus Quantized Region:us Safetensors Sglang Sharded Supertune Tensorflow Vllm

Model Card on HF 🤗: https://huggingface.co/Jiunsong/SuperGLM-5.2-abliterated-NVFP4

SuperGLM 5.2 Abliterated NVFP4 Parameters and Internals

LLM Name	SuperGLM 5.2 Abliterated NVFP4
Repository 🤗	https://huggingface.co/Jiunsong/SuperGLM-5.2-abliterated-NVFP4
Base Model(s)	nvidia/GLM-5.2-NVFP4 nvidia/GLM-5.2-NVFP4
Model Size	381b
Required VRAM	410 GB
Updated	2026-07-21
Maintainer	Jiunsong
Model Type	glm_moe_dsa
Model Files	10.0 GB: 1-of-47 10.0 GB: 2-of-47 10.0 GB: 3-of-47 10.0 GB: 4-of-47 10.0 GB: 5-of-47 10.0 GB: 6-of-47 10.0 GB: 7-of-47 10.0 GB: 8-of-47 10.0 GB: 9-of-47 10.0 GB: 10-of-47 10.0 GB: 11-of-47 10.0 GB: 12-of-47 10.0 GB: 13-of-47 10.0 GB: 14-of-47 10.0 GB: 15-of-47 10.0 GB: 16-of-47 10.0 GB: 17-of-47 10.0 GB: 18-of-47 10.0 GB: 19-of-47 10.0 GB: 20-of-47 10.0 GB: 21-of-47 10.0 GB: 22-of-47 10.0 GB: 23-of-47 10.0 GB: 24-of-47 10.0 GB: 25-of-47 10.0 GB: 26-of-47 10.0 GB: 27-of-47 10.0 GB: 28-of-47 10.0 GB: 29-of-47 10.0 GB: 30-of-47 10.0 GB: 31-of-47 10.0 GB: 32-of-47 10.0 GB: 33-of-47 10.0 GB: 34-of-47 10.0 GB: 35-of-47 10.0 GB: 36-of-47 10.0 GB: 37-of-47 10.0 GB: 38-of-47 10.0 GB: 39-of-47 10.0 GB: 40-of-47 10.0 GB: 41-of-47
Model Architecture	GlmMoeDsaForCausalLM
License	mit
Context Length	1048576
Model Max Length	1048576
Transformers Version	5.11.0
Vocabulary Size	154880

Rank the SuperGLM 5.2 Abliterated NVFP4 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 SuperGLM 5.2 Abliterated NVFP4

Best Alternatives	Context / RAM	Downloads	Likes
GLM 5.2 NVFP4	1024K / 429.9 GB	1356729	276
GLM 5.2 NVFP4	1024K / 429.9 GB	167	3

Note: green Score (e.g. "73.2") means that the model is better than Jiunsong/SuperGLM-5.2-abliterated-NVFP4.