What are the hardware requirements for SuperHY3 Abliterated NVFP4?

SuperHY3 Abliterated NVFP4 requires approximately 66.6 GB of VRAM and supports a context window of 256K tokens. Quantized variants may run on less VRAM; see the Quantized Models section on this page.

Who developed SuperHY3 Abliterated NVFP4 and how large is it?

SuperHY3 Abliterated NVFP4 is developed by Jiunsong, a model with 172b 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 SuperHY3 Abliterated NVFP4?

SuperHY3 Abliterated NVFP4 is hosted on Hugging Face and linked from this page. LLM Explorer also lists quantized variants and similar alternatives if available.

SuperHY3 Abliterated NVFP4 by Jiunsong — VRAM 66.6GB, 256K context

Name: SuperHY3 Abliterated NVFP4
Author: Jiunsong

SuperHY3 Abliterated NVFP4 is an open-source language model by Jiunsong. Features: 172b LLM, VRAM: 66.6GB, Context: 256K, License: apache-2.0, LLM Explorer Score: 0.3.

8-bit Abliterated Base model:finetune:tencent/hy... Base model:tencent/hy3 Compressed-tensors Conversational Endpoints compatible Hunyuan Hy3 Hy v3 Moe Nvfp4 Obliteratus Quantized Region:us Safetensors Sharded Supertune Tensorflow Vllm W4a16

Model Card on HF 🤗: https://huggingface.co/Jiunsong/SuperHY3-abliterated-NVFP4

SuperHY3 Abliterated NVFP4 Parameters and Internals

LLM Name	SuperHY3 Abliterated NVFP4
Repository 🤗	https://huggingface.co/Jiunsong/SuperHY3-abliterated-NVFP4
Base Model(s)	Hy3 tencent/Hy3
Model Size	172b
Required VRAM	66.6 GB
Updated	2026-07-28
Maintainer	Jiunsong
Model Type	hy_v3
Model Files	2.0 GB: 1-of-99 2.0 GB: 2-of-99 2.0 GB: 3-of-99 2.0 GB: 4-of-99 2.0 GB: 5-of-99 1.1 GB: 6-of-99 2.0 GB: 7-of-99 2.0 GB: 8-of-99 2.0 GB: 9-of-99 2.0 GB: 10-of-99 2.0 GB: 11-of-99 1.1 GB: 12-of-99 2.0 GB: 13-of-99 2.0 GB: 14-of-99 2.0 GB: 15-of-99 2.0 GB: 16-of-99 2.0 GB: 17-of-99 1.1 GB: 18-of-99 2.0 GB: 19-of-99 2.0 GB: 20-of-99 2.0 GB: 21-of-99 2.0 GB: 22-of-99 2.0 GB: 23-of-99 1.1 GB: 24-of-99 2.0 GB: 25-of-99 2.0 GB: 26-of-99 2.0 GB: 27-of-99 2.0 GB: 28-of-99 2.0 GB: 29-of-99 1.1 GB: 30-of-99 2.0 GB: 31-of-99 2.0 GB: 32-of-99 2.0 GB: 33-of-99 2.0 GB: 34-of-99 2.0 GB: 35-of-99 1.1 GB: 36-of-99
Model Architecture	HYV3ForCausalLM
License	apache-2.0
Context Length	262144
Model Max Length	262144
Transformers Version	5.6.0
Vocabulary Size	120832

Rank the SuperHY3 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

Quantized Models of the SuperHY3 Abliterated NVFP4

Model	Likes	Downloads	VRAM
SuperHY3 Abliterated MLX 4bit	4	474	169 GB

Best Alternatives to SuperHY3 Abliterated NVFP4

Best Alternatives	Context / RAM	Downloads	Likes
Hy3 NVFP4 W4A16	256K / 79.7 GB	8720	15

Note: green Score (e.g. "73.2") means that the model is better than Jiunsong/SuperHY3-abliterated-NVFP4.