Skip to main content

Run ChindaLLM 4B in OpenWebUI

By the end, ChindaLLM 4B answers in Thai in a browser chat that runs entirely on your own machine.

You need: Docker (Docker Desktop on Windows or macOS), disk space for the OpenWebUI image and the 2.5 GB model, and at least 4 GB of RAM (8 GB recommended). An NVIDIA GPU is optional.

1. Start OpenWebUI​

One container runs OpenWebUI with Ollama inside it.

docker run -d -p 3000:8080 -v ollama:/root/.ollama -v open-webui:/app/backend/data --name open-webui --restart always ghcr.io/open-webui/open-webui:ollama

2. Create the admin account​

Open http://localhost:3000 and sign up; the account is kept on your machine and no data is sent out.

3. Pull the model​

In the model search box, type iapp/chinda-qwen3-4b and click Pull "iapp/chinda-qwen3-4b" from Ollama.com.

Pulling iapp/chinda-qwen3-4b from the OpenWebUI model search

Or from a terminal:

docker exec -it open-webui ollama pull iapp/chinda-qwen3-4b

4. Chat in Thai​

Pick iapp/chinda-qwen3-4b:latest in the model menu and write in Thai.

ChindaLLM 4B answering in Thai in OpenWebUI

For facts, paste the source text into the chat: asked about recent news, statistics or people with no context, a 4-billion-parameter model can make things up.

5. Tune it (optional)​

In the model settings, Temperature 0.7, Top P 0.9 and Top K 40 suit Chinda, and a system prompt sets its role:

คุณเป็นผู้ช่วยเขียนเอกสารมืออาชีพ ตอบด้วยภาษาไทยที่เป็นทางการและสุภาพ
Troubleshooting
  • Chinda is not in the model menu. Check it downloaded with docker exec -it open-webui ollama list, pull it again if it is missing, then refresh the page.
  • The page does not load. Check the container with docker ps -a and docker logs open-webui, then docker restart open-webui.
  • Answers are slow. With a GPU, check nvidia-smi, and if the container was not started with --gpus all, run docker stop open-webui and docker rm open-webui, then the NVIDIA GPU command. Without one, close other applications and lower Max Tokens.
Update, back up and manage models
# Update OpenWebUI, then run the step 1 command again
docker pull ghcr.io/open-webui/open-webui:ollama
docker stop open-webui
docker rm open-webui

# Back up both volumes to the current folder
docker run --rm -v ollama:/source -v $(pwd):/backup alpine tar czf /backup/ollama-backup.tar.gz -C /source .
docker run --rm -v open-webui:/source -v $(pwd):/backup alpine tar czf /backup/openwebui-backup.tar.gz -C /source .

# Pull or remove any other model
docker exec -it open-webui ollama pull <model-name>
docker exec -it open-webui ollama rm <model-name>

The model is Apache 2.0, including commercial use: model page, Hugging Face, Ollama, OpenWebUI on GitHub. To try it without installing, choose ChindaLLM 4B at chindax.iapp.co.th.