Update Dockerfile

Dockerfile

@@ -1,21 +1,35 @@
-FROM ghcr.io/ggml-org/llama.cpp:server
+FROM ghcr.io/ggml-org/llama.cpp:full
 
-RUN apt update && apt install wget -y && rm -rf /var/lib/apt/lists/*
+WORKDIR /app
+
+RUN apt update && apt install -y python3 python3-pip python3-venv wget && rm -rf /var/lib/apt/lists/*
 
 # Create a dedicated directory to organize the models
 RUN mkdir -p /models
 
+# Set up Python virtual environment for huggingface_hub tools
+RUN python3 -m venv /opt/venv
+ENV PATH="/opt/venv/bin:$PATH"
+RUN pip install -U pip huggingface_hub
+
 # --- MODEL 1: Gemma 4 Multimodal + Speculative Setup ---
-RUN wget "https://huggingface.co/unsloth/gemma-4-E2B-it-qat-GGUF/resolve/main/gemma-4-E2B-it-qat-UD-Q4_K_XL.gguf" -O /models/gemma-4-E2B-it-qat-UD-Q4_K_XL.gguf
-RUN wget "https://huggingface.co/unsloth/gemma-4-E2B-it-qat-GGUF/resolve/main/mtp-gemma-4-E2B-it.gguf" -O /models/mtp-gemma-4-E2B-it.gguf
-RUN wget "https://huggingface.co/unsloth/gemma-4-E2B-it-qat-GGUF/resolve/main/mmproj-F16.gguf" -O /models/gemma-4-E2B-it-mmproj.gguf
+RUN wget "https://huggingface.co/unsloth/gemma-4-E2B-it-qat-GGUF/resolve/main/gemma-4-E2B-it-qat-UD-Q4_K_XL.gguf" -O /models/gemma-4-E2B-it-qat-UD-Q4_K_XL.gguf && \
+    wget "https://huggingface.co/unsloth/gemma-4-E2B-it-qat-GGUF/resolve/main/mtp-gemma-4-E2B-it.gguf" -O /models/mtp-gemma-4-E2B-it.gguf && \
+    wget "https://huggingface.co/unsloth/gemma-4-E2B-it-qat-GGUF/resolve/main/mmproj-F16.gguf" -O /models/gemma-4-E2B-it-mmproj.gguf
 
-# --- MODEL 2: Gemma 3 1B Model (Corrected Direct Download Link) ---
+# --- MODEL 2: Gemma 3 1B Model ---
 RUN wget "https://huggingface.co/MaziyarPanahi/gemma-3-1b-it-GGUF/resolve/main/gemma-3-1b-it.Q3_K_M.gguf" -O /models/gemma-3-1b-it.Q3_K_M.gguf
 
+# --- MODEL 3: Gemma 4 26B Multimodal (Added via huggingface_hub) ---
+ARG HF_TOKEN
+RUN python3 -c 'from huggingface_hub import hf_hub_download; \
+    import os; \
+    token = os.getenv("HF_TOKEN") or None; \
+    repo="unsloth/gemma-4-26B-A4B-it-GGUF"; \
+    hf_hub_download(repo_id=repo, filename="gemma-4-26B-A4B-it-UD-Q4_K_XL.gguf", local_dir="/models", token=token); \
+    hf_hub_download(repo_id=repo, filename="mmproj-BF16.gguf", local_dir="/models", token=token)'
 
 # --- CREATE ROUTER CONFIGURATION ---
-# Appending configuration step-by-step to avoid any multiline shell escaping issues
 RUN echo "[gemma-4-vision]" > /models.ini && \
     echo "model = /models/gemma-4-E2B-it-qat-UD-Q4_K_XL.gguf" >> /models.ini && \
     echo "spec-draft-model = /models/mtp-gemma-4-E2B-it.gguf" >> /models.ini && \
@@ -30,8 +44,14 @@ RUN echo "[gemma-4-vision]" > /models.ini && \
     echo "[gemma-3-1b]" >> /models.ini && \
     echo "model = /models/gemma-3-1b-it.Q3_K_M.gguf" >> /models.ini && \
     echo "ctx-size = 4096" >> /models.ini && \
-    echo "flash-attn = on" >> /models.ini
-
+    echo "flash-attn = on" >> /models.ini && \
+    echo "" >> /models.ini && \
+    echo "[gemma-4-26b-vision]" >> /models.ini && \
+    echo "model = /models/gemma-4-26B-A4B-it-UD-Q4_K_XL.gguf" >> /models.ini && \
+    echo "mmproj = /models/mmproj-BF16.gguf" >> /models.ini && \
+    echo "cache-type-k = q8_0" >> /models.ini && \
+    echo "cache-type-v = iq4_nl" >> /models.ini && \
+    echo "ctx-size = 64000" >> /models.ini
 
 # --- START SERVER IN ROUTER MODE ---
 CMD [ \