Create Dockerfile

Dockerfile

@@ -0,0 +1,30 @@
+FROM nvidia/cuda:12.1.1-runtime-ubuntu20.04
+
+# Install basic system deps
+RUN apt-get update && apt-get install -y \
+    git python3 python3-pip ffmpeg libsndfile1 \
+    && rm -rf /var/lib/apt/lists/*
+
+# Install Python deps
+RUN pip install --upgrade pip
+
+# Install nightly vLLM with audio support
+RUN pip install "vllm[audio]" --extra-index-url https://wheels.vllm.ai/nightly
+
+# Optional: mistral_common should come with vllm[audio]
+RUN pip install mistral_common[audio] openai gradio huggingface_hub
+
+# Expose the default Gradio port
+EXPOSE 7860
+
+# Set environment variables
+ENV HF_HUB_ENABLE_HF_TRANSFER=1
+
+# Download and launch model + Gradio client
+COPY app.py .
+
+CMD vllm.serve mistralai/Voxtral-Mini-3B-2507 \
+    --tokenizer_mode mistral \
+    --config_format mistral \
+    --load_format mistral & \
+    python3 app.py