diff --git a/backend/Dockerfile.model_server b/backend/Dockerfile.model_server
index 90ded483122..080d946636a 100644
--- a/backend/Dockerfile.model_server
+++ b/backend/Dockerfile.model_server
@@ -24,15 +24,15 @@ RUN apt-get remove -y --allow-remove-essential perl-base && \
 # Download tokenizers, distilbert for the Danswer model
 # Download model weights
 # Run Nomic to pull in the custom architecture and have it cached locally
-RUN python -c "from transformers import AutoTokenizer; \
-AutoTokenizer.from_pretrained('mixedbread-ai/mxbai-rerank-xsmall-v1'); \
-from huggingface_hub import snapshot_download; \
-snapshot_download(repo_id='danswer/hybrid-intent-token-classifier', revision='v1.0.3'); \
-snapshot_download('mixedbread-ai/mxbai-rerank-xsmall-v1');"
+# RUN python -c "from transformers import AutoTokenizer; \
+# AutoTokenizer.from_pretrained('mixedbread-ai/mxbai-rerank-xsmall-v1'); \
+# from huggingface_hub import snapshot_download; \
+# snapshot_download(repo_id='danswer/hybrid-intent-token-classifier', revision='v1.0.3'); \
+# snapshot_download('mixedbread-ai/mxbai-rerank-xsmall-v1');"
 
 # In case the user has volumes mounted to /root/.cache/huggingface that they've downloaded while
 # running Danswer, don't overwrite it with the built in cache folder
-RUN mv /root/.cache/huggingface /root/.cache/temp_huggingface
+# RUN mv /root/.cache/huggingface /root/.cache/temp_huggingface
 
 WORKDIR /app