FROM ollama/ollama:latest

# Port setup for Hugging Face Spaces
ENV OLLAMA_HOST=0.0.0.0:7860
EXPOSE 7860

# Start server, wait 10s, pull the FAST 1.5B coder model, and keep container alive
ENTRYPOINT ["sh", "-c", "ollama serve & sleep 10 && ollama pull qwen2.5-coder:1.5b && tail -f /dev/null"]