# Usa a imagem oficial do servidor do llama.cpp (Motor ultra leve em C++)
FROM ghcr.io/ggml-org/llama.cpp:server

USER root
RUN apt-get update && apt-get install -y curl && rm -rf /var/lib/apt/lists/*
RUN mkdir -p /models && chmod 777 /models

USER 1000
WORKDIR /models

# BAIXANDO O QWEN 7B (O Rei do Tool Calling Open Source)
RUN curl -L -o model.gguf "https://huggingface.co/Qwen/Qwen2.5-Coder-7B-Instruct-GGUF/resolve/main/qwen2.5-coder-7b-instruct-q4_k_m.gguf"

EXPOSE 7860
ENV LD_LIBRARY_PATH=/app

# START FULL OTIMIZADO: Contexto 8K e 2 Threads para evitar Timeout na CPU gratuita
ENTRYPOINT ["/app/llama-server"]
CMD ["--model", "/models/model.gguf", "--host", "0.0.0.0", "--port", "7860", "-c", "8192", "--threads", "2", "--batch-size", "512"]