What are the hardware requirements for Gaze Attention Qwen3 4B 384?

Gaze Attention Qwen3 4B 384 requires approximately 9.7 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 Gaze Attention Qwen3 4B 384 and how large is it?

Gaze Attention Qwen3 4B 384 is developed by junha1125, a model with 4b 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 Gaze Attention Qwen3 4B 384?

Gaze Attention Qwen3 4B 384 is hosted on Hugging Face and linked from this page. LLM Explorer also lists quantized variants and similar alternatives if available.

Gaze Attention Qwen3 4B 384 by junha1125 — VRAM 9.7GB, 256K context

Name: Gaze Attention Qwen3 4B 384
Author: junha1125

Gaze Attention Qwen3 4B 384 is an open-source language model by junha1125. Features: 4b LLM, VRAM: 9.7GB, Context: 256K, License: apache-2.0, Instruction-Based, LLM Explorer Score: 0.28.

Arxiv:2605.13080 Base model:finetune:qwen/qwen3... Base model:qwen/qwen3-4b-instr... Cambrian Cambrian qwen3 Conversational Dataset:nyu-visionx/cambrian-1... Dataset:nyu-visionx/cambrian-a... En Gaze-attention Image-text-to-text Instruct Multimodal Region:us Safetensors Sharded Tensorflow Vision Vision-language

Model Card on HF 🤗: https://huggingface.co/junha1125/gaze-attention-qwen3-4b-384

Gaze Attention Qwen3 4B 384 Parameters and Internals

LLM Name	Gaze Attention Qwen3 4B 384
Repository 🤗	https://huggingface.co/junha1125/gaze-attention-qwen3-4b-384
Base Model(s)	Qwen3 4B Instruct 2507 google/siglip2-so400m-patch16-384 Qwen/Qwen3-4B-Instruct-2507 google/siglip2-so400m-patch16-384
Model Size	4b
Required VRAM	9.7 GB
Updated	2026-10-05
Maintainer	junha1125
Model Type	cambrian_qwen3
Instruction-Based	Yes
Model Files	5.0 GB: 1-of-2 4.7 GB: 2-of-2
Supported Languages	en
Model Architecture	CambrianQwen3ForCausalLM
License	apache-2.0
Context Length	262144
Model Max Length	262144
Transformers Version	4.51.0
Tokenizer Class	Qwen2Tokenizer
Padding Token	<\|endoftext\|>
Vocabulary Size	151673
Torch Data Type	float16
Errors	replace

Rank the Gaze Attention Qwen3 4B 384 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 Gaze Attention Qwen3 4B 384

Best Alternatives	Context / RAM	Downloads	Likes
Gaze Attention Qwen3 4B 512	256K / 9.7 GB	21	1
...tention Qwen3 4B 384 Onevision	256K / 9.7 GB	20	1