Qwen 3.8 27B (IQ4_XS) - Optimized for 16GB VRAM

ollama run VladimirGav/qwen3.8-27B-14GB-IQ4

curl http://localhost:11434/api/chat \
  -d '{
    "model": "VladimirGav/qwen3.8-27B-14GB-IQ4",
    "messages": [{"role": "user", "content": "Hello!"}]
  }'

from ollama import chat

response = chat(
    model='VladimirGav/qwen3.8-27B-14GB-IQ4',
    messages=[{'role': 'user', 'content': 'Hello!'}],
)
print(response.message.content)

import ollama from 'ollama'

const response = await ollama.chat({
  model: 'VladimirGav/qwen3.8-27B-14GB-IQ4',
  messages: [{role: 'user', content: 'Hello!'}],
})
console.log(response.message.content)

Models

View all →

Name

1 model

Size / Usage

Context

Input

qwen3.8-27B-14GB-IQ4:latest

14GB · 256K context window · Text · 2 days ago

qwen3.8-27B-14GB-IQ4:latest

14GB

256K

Text

Readme

Qwen 3.8 27B (IQ4) - Optimized for 16GB VRAM

This is a highly optimized version of Alibaba Qwen 3.8 27B, specifically tailored to run on GPUs with 16GB of VRAM. It uses the advanced IQ4 (Importance Matrix) quantization to maintain high intelligence while fitting safely into the memory footprint.

🚀 Key Features

VRAM Efficient: The model weights occupy ~14GB, leaving a safe ~2GB buffer for context (KV Cache) and system overhead on a 16GB card.

💻 Target Hardware

Perfectly fits 16GB VRAM GPUs: * NVIDIA RTX 5060 Ti (16GB) * NVIDIA RTX 4070 Ti Super (16GB) * NVIDIA RTX 3080 (16GB version) * NVIDIA RTX 4080 / 4080 Super * NVIDIA RTX 5000 / A4000 (16GB)

🛠 How to Use

Simply run the following command in your terminal:

ollama run VladimirGav/qwen3.8-27B-14GB-IQ4