From daa454c220bcba2cf66b8069cdd4e00f38bbddf1 Mon Sep 17 00:00:00 2001 From: George Panchuk Date: Thu, 24 Sep 2026 22:49:55 +0700 Subject: [PATCH 1/2] deprecate: drop fallback url to a google bucket --- fastembed/common/model_management.py | 1 - fastembed/text/onnx_embedding.py | 4 ---- fastembed/text/pooled_embedding.py | 1 - fastembed/text/pooled_normalized_embedding.py | 1 - 4 files changed, 7 deletions(-) diff --git a/fastembed/common/model_management.py b/fastembed/common/model_management.py index d044559de..e4ef25e0a 100644 --- a/fastembed/common/model_management.py +++ b/fastembed/common/model_management.py @@ -440,7 +440,6 @@ def download_model(cls, model: T, cache_dir: str, retries: int = 3, **kwargs: An "description": "Base English model, v1.5", "size_in_GB": 0.44, "sources": { - "url": "https://storage.googleapis.com/qdrant-fastembed/fast-bge-base-en-v1.5.tar.gz", "hf": "qdrant/bge-base-en-v1.5-onnx-q", } } diff --git a/fastembed/text/onnx_embedding.py b/fastembed/text/onnx_embedding.py index ff527fc6f..6d14e8459 100644 --- a/fastembed/text/onnx_embedding.py +++ b/fastembed/text/onnx_embedding.py @@ -19,7 +19,6 @@ size_in_GB=0.42, sources=ModelSource( hf="Qdrant/fast-bge-base-en", - url="https://storage.googleapis.com/qdrant-fastembed/fast-bge-base-en.tar.gz", _deprecated_tar_struct=True, ), model_file="model_optimized.onnx", @@ -35,7 +34,6 @@ size_in_GB=0.21, sources=ModelSource( hf="Qdrant/bge-base-en-v1.5-onnx-Q", - url="https://storage.googleapis.com/qdrant-fastembed/fast-bge-base-en-v1.5.tar.gz", _deprecated_tar_struct=True, ), model_file="model_optimized.onnx", @@ -63,7 +61,6 @@ size_in_GB=0.13, sources=ModelSource( hf="Qdrant/bge-small-en", - url="https://storage.googleapis.com/qdrant-fastembed/BAAI-bge-small-en.tar.gz", _deprecated_tar_struct=True, ), model_file="model_optimized.onnx", @@ -91,7 +88,6 @@ size_in_GB=0.09, sources=ModelSource( hf="Qdrant/bge-small-zh-v1.5", - url="https://storage.googleapis.com/qdrant-fastembed/fast-bge-small-zh-v1.5.tar.gz", _deprecated_tar_struct=True, ), model_file="model_optimized.onnx", diff --git a/fastembed/text/pooled_embedding.py b/fastembed/text/pooled_embedding.py index f3fe607e9..a99c7c42b 100644 --- a/fastembed/text/pooled_embedding.py +++ b/fastembed/text/pooled_embedding.py @@ -81,7 +81,6 @@ size_in_GB=2.24, sources=ModelSource( hf="qdrant/multilingual-e5-large-onnx", - url="https://storage.googleapis.com/qdrant-fastembed/fast-multilingual-e5-large.tar.gz", _deprecated_tar_struct=True, ), model_file="model.onnx", diff --git a/fastembed/text/pooled_normalized_embedding.py b/fastembed/text/pooled_normalized_embedding.py index 44511d267..637096023 100644 --- a/fastembed/text/pooled_normalized_embedding.py +++ b/fastembed/text/pooled_normalized_embedding.py @@ -19,7 +19,6 @@ license="apache-2.0", size_in_GB=0.09, sources=ModelSource( - url="https://storage.googleapis.com/qdrant-fastembed/sentence-transformers-all-MiniLM-L6-v2.tar.gz", hf="qdrant/all-MiniLM-L6-v2-onnx", _deprecated_tar_struct=True, ), From 53c62354507a97f536e925761027cab8f2f40a51 Mon Sep 17 00:00:00 2001 From: George Panchuk Date: Fri, 25 Sep 2026 00:08:38 +0700 Subject: [PATCH 2/2] deprecation: add a deprecation warning when a model is loaded using cached gcs data --- fastembed/common/model_management.py | 13 +++++++++++++ 1 file changed, 13 insertions(+) diff --git a/fastembed/common/model_management.py b/fastembed/common/model_management.py index e4ef25e0a..c53d72485 100644 --- a/fastembed/common/model_management.py +++ b/fastembed/common/model_management.py @@ -5,6 +5,7 @@ import shutil import tarfile import tempfile +import warnings import contextlib from copy import deepcopy from pathlib import Path, PureWindowsPath @@ -374,6 +375,18 @@ def retrieve_model_gcs( # check if the model_dir and the model files are both present for macOS if model_dir.exists() and len(list(model_dir.glob("*"))) > 0: + if deprecated_tar_struct: + # No built-in model is served from the bucket anymore, only copies of it remain. + # stacklevel points at the caller of TextEmbedding(...), the path that gets here. + warnings.warn( + f"Loading {model_name} from {model_dir}, a copy downloaded from Google Cloud " + "Storage by an older fastembed version. Support for such copies is deprecated " + "and will be removed in a future release. To switch to Hugging Face, load the " + "model once with network access (without `local_files_only=True` or " + f"`HF_HUB_OFFLINE=1`), then delete {model_dir}.", + FutureWarning, + stacklevel=5, + ) return model_dir if local_files_only: