LisyNeko/ qwen3.8-9b-coder:latest

2,358 Downloads Updated 1 week ago

vision tools thinking

ollama run LisyNeko/qwen3.8-9b-coder

curl http://localhost:11434/api/chat \
  -d '{
    "model": "LisyNeko/qwen3.8-9b-coder",
    "messages": [{"role": "user", "content": "Hello!"}]
  }'

from ollama import chat

response = chat(
    model='LisyNeko/qwen3.8-9b-coder',
    messages=[{'role': 'user', 'content': 'Hello!'}],
)
print(response.message.content)

import ollama from 'ollama'

const response = await ollama.chat({
  model: 'LisyNeko/qwen3.8-9b-coder',
  messages: [{role: 'user', content: 'Hello!'}],
})
console.log(response.message.content)

Details

Updated 1 week ago

1 week ago

963a6838c512 · 6.7GB ·

model

archqwen35

parameters9.2B

quantizationQ4_K_M

5.8GB

projector

archclip

parameters456M

quantizationBF16

922MB

system

You are Qwen3.8-9B, an expert programming assistant. When asked about your identity, model version,

274B

params

{ "repeat_penalty": 1.05, "temperature": 0.7, "top_k": 20, "top_p": 0.8 }

65B

template

13B

Readme

qwen3.8-9b-coder

Custom coding-focused LLM. Deployed on Ollama as: LisyNeko/qwen3.8-9b-coder

Specs

Size: 6.7GB
Context window: 256K
Input: Text, Image (vision)

Base Model

empero-ai/Qwen3.8-9B (Hugging Face)

Architecture: Qwen3.5-9B
Distillation source: Qwen3.8 2.4T-A95B → full-parameter distilled into 9B
Original state: text-only (no vision support)

Improvements

The original (empero-ai/Qwen3.8-9B) did not support image input. Building on it, the following was added: - Vision (image input) capability - Coding-focused fine-tuning

→ Result: a custom model combining Qwen3.5-9B architecture + Qwen3.8 knowledge distillation + vision integration + coder specialization