deprecate: drop fallback url to a google bucket (#726 )

* deprecate: drop fallback url to a google bucket * deprecation: add a deprecation warning when a model is loaded using cached gcs data
fix: extend the pool of network errors caught during download (#727 )
2026-09-24 15:07:50 -05:00 · 2026-09-25 02:23:15 +07:00 · 2026-09-25 02:14:12 +07:00 · 2026-09-24 16:40:52 +07:00 · 2026-09-23 02:44:32 +07:00
13 changed files with 121 additions and 55 deletions
@@ -1,11 +1,10 @@
 name: Tests
-run-name: Tests (gpu)

 on:
  pull_request:
    branches: [ master, main, gpu ]
  workflow_dispatch:
-
+  

 env:
  CARGO_TERM_COLOR: always
@@ -22,6 +21,22 @@ jobs:
          - '3.13.x'
        os:
          - ubuntu-latest
+          - macos-latest
+          - windows-latest
+        exclude:
+          # Exclude 3.10–3.12 for macOS and Windows
+          - os: macos-latest
+            python-version: '3.10.x'
+          - os: macos-latest
+            python-version: '3.11.x'
+          - os: macos-latest
+            python-version: '3.12.x'
+          - os: windows-latest
+            python-version: '3.10.x'
+          - os: windows-latest
+            python-version: '3.11.x'
+          - os: windows-latest
+            python-version: '3.12.x'

    runs-on: ${{ matrix.os }}

@@ -12,8 +12,8 @@ This distribution includes the following Jina AI models, each with its respectiv

 These models are developed by Jina (https://jina.ai/) and are subject to Jina AI's licensing terms.

-This distribution includes the following Google models, each with its respective license:
- vidore/colpali-v1.3
+This distribution includes the following Gemma-based models, each with its respective license:
+- Qdrant/colpali-v1.3-fp16
    - License: gemma
 - google/embeddinggemma-300m
    - License: gemma
@@ -5,6 +5,7 @@ import json
 import shutil
 import tarfile
 import tempfile
+import warnings
 import contextlib
 from copy import deepcopy
 from pathlib import Path, PureWindowsPath
@@ -27,6 +28,27 @@ T = TypeVar("T", bound=BaseModelDescription)
 _DOWNLOAD_CHUNK_SIZE = 256 * 1024


+def _hf_transport_errors() -> tuple[type[Exception], ...]:
+    """Network errors of huggingface_hub's HTTP library that aren't OSError.
+
+    A refused connection, a DNS failure or a timeout raises an OSError in huggingface_hub 0.x,
+    which is built on requests, but a TransportError in 1.x (httpx) and 2.x (httpx2).
+    """
+    try:
+        # huggingface_hub>=1.30 re-exports whichever of httpx and httpx2 it's built on.
+        from huggingface_hub.utils import httpx
+    except ImportError:
+        try:
+            import httpx  # huggingface_hub 1.0 to 1.29
+        except ImportError:  # huggingface_hub 0.x
+            return ()
+    return (httpx.TransportError,)
+
+
+# Errors from an HF download that download_model handles by falling back to url and retrying.
+_HF_DOWNLOAD_ERRORS = (OSError, RepositoryNotFoundError, ValueError) + _hf_transport_errors()
+
+
 class ModelManagement(Generic[T]):
    METADATA_FILE = "files_metadata.json"

@@ -374,6 +396,18 @@ class ModelManagement(Generic[T]):

        # 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:
@@ -440,7 +474,6 @@ class ModelManagement(Generic[T]):
                    "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",
                    }
                }
@@ -502,7 +535,7 @@ class ModelManagement(Generic[T]):
                            **kwargs,
                        )
                    )
-                except (EnvironmentError, RepositoryNotFoundError, ValueError) as e:
+                except _HF_DOWNLOAD_ERRORS as e:
                    if not local_files_only:
                        logger.error(
                            f"Could not download model from HuggingFace: {e} "
@@ -33,7 +33,7 @@ supported_onnx_models: list[DenseModelDescription] = [
        model="Qdrant/Unicom-ViT-B-16",
        dim=768,
        description="Image embeddings (more detailed than Unicom-ViT-B-32), Multimodal (text&image), 2023 year",
-        license="apache-2.0",
+        license="mit",
        size_in_GB=0.82,
        sources=ModelSource(hf="Qdrant/Unicom-ViT-B-16"),
        model_file="model.onnx",
@@ -42,7 +42,7 @@ supported_onnx_models: list[DenseModelDescription] = [
        model="Qdrant/Unicom-ViT-B-32",
        dim=512,
        description="Image embeddings, Multimodal (text&image), 2023 year",
-        license="apache-2.0",
+        license="mit",
        size_in_GB=0.48,
        sources=ModelSource(hf="Qdrant/Unicom-ViT-B-32"),
        model_file="model.onnx",
@@ -221,6 +221,8 @@ class Colbert(LateInteractionTextEmbeddingBase, OnnxTextModel[NumpyArray]):
        self.query_tokenizer, _ = load_tokenizer(model_dir=self._model_dir)

        assert self.tokenizer is not None
+        # load_tokenizer always enables both
+        assert self.tokenizer.padding is not None and self.tokenizer.truncation is not None
        self.mask_token_id = self.special_token_to_id[self.MASK_TOKEN]
        self.pad_token_id = self.tokenizer.padding["pad_id"]
        self.skip_list = {
@@ -22,7 +22,7 @@ supported_colpali_models: list[DenseModelDescription] = [
        model="Qdrant/colpali-v1.3-fp16",
        dim=128,
        description="Text embeddings, Multimodal (text&image), English, 50 tokens query length truncation, 2024.",
-        license="mit",
+        license="gemma",
        size_in_GB=6.5,
        sources=ModelSource(hf="Qdrant/colpali-v1.3-fp16"),
        additional_files=["model.onnx_data"],
@@ -33,7 +33,7 @@ class LateInteractionMultimodalEmbedding(LateInteractionMultimodalEmbeddingBase)
                         "model": "Qdrant/colpali-v1.3-fp16",
                         "dim": 128,
                         "description": "Text embeddings, Unimodal (text), Aligned to image latent space, ColBERT-compatible, 512 tokens max, 2024.",
-                         "license": "mit",
+                         "license": "gemma",
                         "size_in_GB": 6.06,
                         "sources": {
                            "hf": "Qdrant/colpali-v1.3-fp16",
@@ -25,11 +25,19 @@ class VocabTokenizer(VocabTokenizerBase):
        return np.array(self.tokenizer.encode(sentence).ids)

    def convert_ids_to_tokens(self, token_ids: NumpyArray) -> list[str]:
-        return [self.tokenizer.id_to_token(token_id) for token_id in token_ids]
+        tokens = []
+        for token_id in token_ids:
+            token = self.tokenizer.id_to_token(token_id)
+            if token is None:
+                raise ValueError(f"Token id {token_id} is not in the vocabulary")
+            tokens.append(token)
+        return tokens


 class VocabResolver:
-    def __init__(self, tokenizer: VocabTokenizerBase, stopwords: set[str], stemmer: SnowballStemmer):
+    def __init__(
+        self, tokenizer: VocabTokenizerBase, stopwords: set[str], stemmer: SnowballStemmer
+    ):
        # Word to id mapping
        self.vocab: dict[str, int] = {}
        # Id to word mapping
@@ -199,4 +207,3 @@ class VocabResolver:
            else:
                counts[vocab_id] += 1
        return token_ids, counts, oov_count, forms
-
@@ -19,7 +19,6 @@ supported_onnx_models: list[DenseModelDescription] = [
        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 @@ supported_onnx_models: list[DenseModelDescription] = [
        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 @@ supported_onnx_models: list[DenseModelDescription] = [
        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 @@ supported_onnx_models: list[DenseModelDescription] = [
        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",
@@ -81,7 +81,6 @@ supported_pooled_models: list[DenseModelDescription] = [
        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",
@@ -19,7 +19,6 @@ supported_pooled_normalized_models: list[DenseModelDescription] = [
        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,
        ),
@@ -2761,21 +2761,34 @@ typing_extensions = ">=4.7.1"
 reference = ["Pillow (>=12.2.0)"]

 [[package]]
-name = "onnxruntime-gpu"
+name = "onnxruntime"
 version = "1.23.2"
 description = "ONNX Runtime is a runtime accelerator for Machine Learning models"
 optional = false
 python-versions = ">=3.10"
 files = [
-    {file = "onnxruntime_gpu-1.23.2-cp310-cp310-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:18de50c6c8eea50acc405ea13d299aec593e46478d7a22cd32cdbbdf7c42899d"},
-    {file = "onnxruntime_gpu-1.23.2-cp310-cp310-win_amd64.whl", hash = "sha256:deba091e15357355aa836fd64c6c4ac97dd0c4609c38b08a69675073ea46b321"},
-    {file = "onnxruntime_gpu-1.23.2-cp311-cp311-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:d76d1ac7a479ecc3ac54482eea4ba3b10d68e888a0f8b5f420f0bdf82c5eec59"},
-    {file = "onnxruntime_gpu-1.23.2-cp311-cp311-win_amd64.whl", hash = "sha256:054282614c2fc9a4a27d74242afbae706a410f1f63cc35bc72f99709029a5ba4"},
-    {file = "onnxruntime_gpu-1.23.2-cp312-cp312-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:4f2d1f720685d729b5258ec1b36dee1de381b8898189908c98cbeecdb2f2b5c2"},
-    {file = "onnxruntime_gpu-1.23.2-cp312-cp312-win_amd64.whl", hash = "sha256:fe925a84b00e291e0ad3fac29bfd8f8e06112abc760cdc82cb711b4f3935bd95"},
-    {file = "onnxruntime_gpu-1.23.2-cp313-cp313-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:1e8f75af5da07329d0c3a5006087f4051d8abd133b4be7c9bae8cdab7bea4c26"},
-    {file = "onnxruntime_gpu-1.23.2-cp313-cp313-win_amd64.whl", hash = "sha256:7f1b3f49e5e126b99e23ec86b4203db41c2a911f6165f7624f2bc8267aaca767"},
-    {file = "onnxruntime_gpu-1.23.2-cp313-cp313t-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:20959cd4ae358aab6579ab9123284a7b1498f7d51ec291d429a5edc26511306f"},
+    {file = "onnxruntime-1.23.2-cp310-cp310-macosx_13_0_arm64.whl", hash = "sha256:a7730122afe186a784660f6ec5807138bf9d792fa1df76556b27307ea9ebcbe3"},
+    {file = "onnxruntime-1.23.2-cp310-cp310-macosx_13_0_x86_64.whl", hash = "sha256:b28740f4ecef1738ea8f807461dd541b8287d5650b5be33bca7b474e3cbd1f36"},
+    {file = "onnxruntime-1.23.2-cp310-cp310-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:8f7d1fe034090a1e371b7f3ca9d3ccae2fabae8c1d8844fb7371d1ea38e8e8d2"},
+    {file = "onnxruntime-1.23.2-cp310-cp310-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:4ca88747e708e5c67337b0f65eed4b7d0dd70d22ac332038c9fc4635760018f7"},
+    {file = "onnxruntime-1.23.2-cp310-cp310-win_amd64.whl", hash = "sha256:0be6a37a45e6719db5120e9986fcd30ea205ac8103fd1fb74b6c33348327a0cc"},
+    {file = "onnxruntime-1.23.2-cp311-cp311-macosx_13_0_arm64.whl", hash = "sha256:6f91d2c9b0965e86827a5ba01531d5b669770b01775b23199565d6c1f136616c"},
+    {file = "onnxruntime-1.23.2-cp311-cp311-macosx_13_0_x86_64.whl", hash = "sha256:87d8b6eaf0fbeb6835a60a4265fde7a3b60157cf1b2764773ac47237b4d48612"},
+    {file = "onnxruntime-1.23.2-cp311-cp311-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:bbfd2fca76c855317568c1b36a885ddea2272c13cb0e395002c402f2360429a6"},
+    {file = "onnxruntime-1.23.2-cp311-cp311-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:da44b99206e77734c5819aa2142c69e64f3b46edc3bd314f6a45a932defc0b3e"},
+    {file = "onnxruntime-1.23.2-cp311-cp311-win_amd64.whl", hash = "sha256:902c756d8b633ce0dedd889b7c08459433fbcf35e9c38d1c03ddc020f0648c6e"},
+    {file = "onnxruntime-1.23.2-cp312-cp312-macosx_13_0_arm64.whl", hash = "sha256:b8f029a6b98d3cf5be564d52802bb50a8489ab73409fa9db0bf583eabb7c2321"},
+    {file = "onnxruntime-1.23.2-cp312-cp312-macosx_13_0_x86_64.whl", hash = "sha256:218295a8acae83905f6f1aed8cacb8e3eb3bd7513a13fe4ba3b2664a19fc4a6b"},
+    {file = "onnxruntime-1.23.2-cp312-cp312-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:76ff670550dc23e58ea9bc53b5149b99a44e63b34b524f7b8547469aaa0dcb8c"},
+    {file = "onnxruntime-1.23.2-cp312-cp312-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:0f9b4ae77f8e3c9bee50c27bc1beede83f786fe1d52e99ac85aa8d65a01e9b77"},
+    {file = "onnxruntime-1.23.2-cp312-cp312-win_amd64.whl", hash = "sha256:25de5214923ce941a3523739d34a520aac30f21e631de53bba9174dc9c004435"},
+    {file = "onnxruntime-1.23.2-cp313-cp313-macosx_13_0_arm64.whl", hash = "sha256:2ff531ad8496281b4297f32b83b01cdd719617e2351ffe0dba5684fb283afa1f"},
+    {file = "onnxruntime-1.23.2-cp313-cp313-macosx_13_0_x86_64.whl", hash = "sha256:162f4ca894ec3de1a6fd53589e511e06ecdc3ff646849b62a9da7489dee9ce95"},
+    {file = "onnxruntime-1.23.2-cp313-cp313-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:45d127d6e1e9b99d1ebeae9bcd8f98617a812f53f46699eafeb976275744826b"},
+    {file = "onnxruntime-1.23.2-cp313-cp313-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:8bace4e0d46480fbeeb7bbe1ffe1f080e6663a42d1086ff95c1551f2d39e7872"},
+    {file = "onnxruntime-1.23.2-cp313-cp313-win_amd64.whl", hash = "sha256:1f9cc0a55349c584f083c1c076e611a7c35d5b867d5d6e6d6c823bf821978088"},
+    {file = "onnxruntime-1.23.2-cp313-cp313t-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:9d2385e774f46ac38f02b3a91a91e30263d41b2f1f4f26ae34805b2a9ddef466"},
+    {file = "onnxruntime-1.23.2-cp313-cp313t-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:e2b9233c4947907fd1818d0e581c049c41ccc39b2856cc942ff6d26317cee145"},
 ]

 [package.dependencies]
@@ -2786,33 +2799,37 @@ packaging = "*"
 protobuf = "*"
 sympy = "*"

-[package.extras]
-cuda = ["nvidia-cuda-nvrtc-cu12 (>=12.0,<13.0)", "nvidia-cuda-runtime-cu12 (>=12.0,<13.0)", "nvidia-cufft-cu12 (>=11.0,<12.0)", "nvidia-curand-cu12 (>=10.0,<11.0)"]
-cudnn = ["nvidia-cudnn-cu12 (>=9.0,<10.0)"]
-
 [[package]]
-name = "onnxruntime-gpu"
+name = "onnxruntime"
 version = "1.30.0"
 description = "ONNX Runtime is a runtime accelerator for Machine Learning models"
 optional = false
 python-versions = ">=3.11"
 files = [
-    {file = "onnxruntime_gpu-1.30.0-cp311-cp311-manylinux_2_28_x86_64.whl", hash = "sha256:4fbf0c5d7b5fb99266c2e8a802e3e64af68bdc383e0f3c7349b985f1087eedaa"},
-    {file = "onnxruntime_gpu-1.30.0-cp311-cp311-manylinux_2_34_aarch64.whl", hash = "sha256:6c70b83afd6bfcc34b4b58d6a41f8c439c96811bbb7fe9fb1d2c2c9dd9c881b5"},
-    {file = "onnxruntime_gpu-1.30.0-cp311-cp311-win_amd64.whl", hash = "sha256:fea66ebd743717de6f254955293b5f978825be7867c87a99c4a0ac1fe6f1d763"},
-    {file = "onnxruntime_gpu-1.30.0-cp312-cp312-manylinux_2_28_x86_64.whl", hash = "sha256:6acc0f4b3d5ac5f2984432faa59dd8fcce6775f8b43a7140afa0f322c7a703b1"},
-    {file = "onnxruntime_gpu-1.30.0-cp312-cp312-manylinux_2_34_aarch64.whl", hash = "sha256:12af352b24145b3480794fcf327d313a240550e772a07fb182390872c1a642fc"},
-    {file = "onnxruntime_gpu-1.30.0-cp312-cp312-win_amd64.whl", hash = "sha256:882da66f936e1f560b155118c5953dc45268cd92893abb8d31a940f4d1effb36"},
-    {file = "onnxruntime_gpu-1.30.0-cp313-cp313-manylinux_2_28_x86_64.whl", hash = "sha256:9aba23f903f434cc55d851aeb801afee19e287c202f4ab8f59d02f04caa90aed"},
-    {file = "onnxruntime_gpu-1.30.0-cp313-cp313-manylinux_2_34_aarch64.whl", hash = "sha256:0c68878729527a28202af6c1b7ebc2fc5a687e9f6e02bfb8796e39d1fe775fd7"},
-    {file = "onnxruntime_gpu-1.30.0-cp313-cp313-win_amd64.whl", hash = "sha256:cf494da233f3fc02bbeb2a3178e3ace7361b6b4fef7031b19c1e5aed5e9328ce"},
-    {file = "onnxruntime_gpu-1.30.0-cp313-cp313t-manylinux_2_28_x86_64.whl", hash = "sha256:675d519cab613bc42685ecb998f8e8e8c6d73143496681650d30ef14fedeade7"},
-    {file = "onnxruntime_gpu-1.30.0-cp313-cp313t-manylinux_2_34_aarch64.whl", hash = "sha256:c6df4b62538f98482dd5cebc91cba668c4b82156e872517c314a383c904a191d"},
-    {file = "onnxruntime_gpu-1.30.0-cp314-cp314-manylinux_2_28_x86_64.whl", hash = "sha256:5aacc98f394a13f4891c0a07e60479d8fb0b1ea6f3328b6b38cfc88f9dfb05ea"},
-    {file = "onnxruntime_gpu-1.30.0-cp314-cp314-manylinux_2_34_aarch64.whl", hash = "sha256:96cb15cff2f0dffaef3ae636560145d41be47bb73e3230113f2e946ca576014b"},
-    {file = "onnxruntime_gpu-1.30.0-cp314-cp314-win_amd64.whl", hash = "sha256:794fc118cdbf340ed02b44c7527e557ea51746002d48105d493ea41f0c9a60d0"},
-    {file = "onnxruntime_gpu-1.30.0-cp314-cp314t-manylinux_2_28_x86_64.whl", hash = "sha256:a4dfbe80d4167ff5bc269f266778ee6d0eb771fd9f3c0c4d83757e7ae42fe3c8"},
-    {file = "onnxruntime_gpu-1.30.0-cp314-cp314t-manylinux_2_34_aarch64.whl", hash = "sha256:13d4cf97120882739da8572ba26e2a0f39caed4c69749a147d8f1a77f74a4690"},
+    {file = "onnxruntime-1.30.0-cp311-cp311-macosx_14_0_arm64.whl", hash = "sha256:f36c87ec504f4f328e968ee322e4ba9682abe66280b3425625afcbed5333fed7"},
+    {file = "onnxruntime-1.30.0-cp311-cp311-manylinux_2_28_aarch64.whl", hash = "sha256:bac3aee7b40acd805659fd81a2acd6096312d5b0f014c97ddfe75fb4c1644627"},
+    {file = "onnxruntime-1.30.0-cp311-cp311-manylinux_2_28_x86_64.whl", hash = "sha256:fd54b314ea385bcecac69ab431f020ba503e3878dad4ebb645fec5a24b041242"},
+    {file = "onnxruntime-1.30.0-cp311-cp311-win_amd64.whl", hash = "sha256:0edd0145a6e3fce8a1276491dc784d615e3c58bcb952c9b4e5c876d5c6a12ad7"},
+    {file = "onnxruntime-1.30.0-cp311-cp311-win_arm64.whl", hash = "sha256:16506cfa5d218850f2b46e0edce1b952121f2f84f8e0074dafe15f4787f0c8fb"},
+    {file = "onnxruntime-1.30.0-cp312-cp312-macosx_14_0_arm64.whl", hash = "sha256:001ed726c9bd5e2bc92faade7d37d889e9606a350b7d5529f0227df2e3bb57fd"},
+    {file = "onnxruntime-1.30.0-cp312-cp312-manylinux_2_28_aarch64.whl", hash = "sha256:6c32a000d5139a38ba9349030b0032e3331acb559d596b22738d9d2b343a2b83"},
+    {file = "onnxruntime-1.30.0-cp312-cp312-manylinux_2_28_x86_64.whl", hash = "sha256:fa688e7891a6aa206636fe7372e27ee75fd17713289f6b4fc7b190e0a7de9328"},
+    {file = "onnxruntime-1.30.0-cp312-cp312-win_amd64.whl", hash = "sha256:f3501472571f1b1eee50e017851e7929f5ea37312d2d8c2494a19e8fc58b4a38"},
+    {file = "onnxruntime-1.30.0-cp312-cp312-win_arm64.whl", hash = "sha256:dc4c706f1935ebb62356e6a095b047859badd854482c40560888e95c328ed262"},
+    {file = "onnxruntime-1.30.0-cp313-cp313-macosx_14_0_arm64.whl", hash = "sha256:05e4fc41711d1f4abd19a9124b5be7a65a506cb2670a7f14b16162ef13c58134"},
+    {file = "onnxruntime-1.30.0-cp313-cp313-manylinux_2_28_aarch64.whl", hash = "sha256:5327cf6aa15a02bad805fac8bd6882a62571e8b72f6f2938a8f37e6bd1966ce9"},
+    {file = "onnxruntime-1.30.0-cp313-cp313-manylinux_2_28_x86_64.whl", hash = "sha256:86f940afc801ea9681a4da8af84fbe95e1d9ea7d80903952cc1bfad54faad38f"},
+    {file = "onnxruntime-1.30.0-cp313-cp313-win_amd64.whl", hash = "sha256:4b63041bd623a9a9ac5e353948436c6fa7f43edd12d6b4a4ebc340bca959ba93"},
+    {file = "onnxruntime-1.30.0-cp313-cp313-win_arm64.whl", hash = "sha256:c389b6887fc95e0fcb80e89b156bc2cb18e662c29df55e9326fe64140e7d7b4f"},
+    {file = "onnxruntime-1.30.0-cp313-cp313t-manylinux_2_28_aarch64.whl", hash = "sha256:e5224ba2b00284cb1c48b3edcd303c109245d66df9ec1b861858fd6de672a38e"},
+    {file = "onnxruntime-1.30.0-cp313-cp313t-manylinux_2_28_x86_64.whl", hash = "sha256:3f9e002417f1e3bbb31ed43dafa4f22ca1b2b68832244fdee192fcaa1ae19bca"},
+    {file = "onnxruntime-1.30.0-cp314-cp314-macosx_14_0_arm64.whl", hash = "sha256:8b6169c16a48429890d2f4a0c774ebf54dfe9066a998514aad0518a16d398547"},
+    {file = "onnxruntime-1.30.0-cp314-cp314-manylinux_2_28_aarch64.whl", hash = "sha256:d2184fddb6798136e7c478244391ca82443f5c757f59f15bb9e5ad2da5e03175"},
+    {file = "onnxruntime-1.30.0-cp314-cp314-manylinux_2_28_x86_64.whl", hash = "sha256:8b611d24db2954545ce6bd9acd4670183cb368e4642450de7a9ab6474eb374ec"},
+    {file = "onnxruntime-1.30.0-cp314-cp314-win_amd64.whl", hash = "sha256:3bdd1752a8502ac1a7ccc6e878d16db6943df574d54ed7f01058a6806ae05be4"},
+    {file = "onnxruntime-1.30.0-cp314-cp314-win_arm64.whl", hash = "sha256:83d543843cbd352cfa9996a6c7b92f8a480f114a18f95ff2a1acaf6921e85b6d"},
+    {file = "onnxruntime-1.30.0-cp314-cp314t-manylinux_2_28_aarch64.whl", hash = "sha256:265de607ba6f9814264e1d5d413fa7069d70f48e3049b5e460b3b04bdfcef294"},
+    {file = "onnxruntime-1.30.0-cp314-cp314t-manylinux_2_28_x86_64.whl", hash = "sha256:67ad7f03433b6462c627d0f555dece80e6a26bc71e8542ced35cebd32142d1b7"},
 ]

 [package.dependencies]
@@ -2822,8 +2839,6 @@ packaging = "*"
 protobuf = ">=4.25.8"

 [package.extras]
-cuda = ["nvidia-cuda-nvrtc (>=13.0,<14.0)", "nvidia-cuda-runtime (>=13.0,<14.0)", "nvidia-cufft (>=12.0,<13.0)", "nvidia-curand (>=10.0,<11.0)"]
-cudnn = ["nvidia-cudnn-cu13 (>=9.0,<10.0)"]
 quantization = ["ml_dtypes"]
 symbolic = ["sympy"]

@@ -4278,4 +4293,4 @@ dev = ["black (>=19.3b0)", "pytest (>=4.6.2)"]
 [metadata]
 lock-version = "2.0"
 python-versions = ">=3.10.0"
-content-hash = "e037421f78383737d79ff2c222fcd317cf9a5754261da077fca1d3fff2f36d0b"
+content-hash = "964982e7dbb361b896623936c15a28dd49c1b585b3b15dd6c712f703051e5aca"
@@ -1,5 +1,5 @@
 [tool.poetry]
-name = "fastembed-gpu"
+name = "fastembed"
 version = "0.8.1"
 description = "Fast, light, accurate library built for retrieval embedding generation"
 authors = ["Qdrant Team <info@qdrant.tech>", "NirantK <nirant.bits@gmail.com>"]
@@ -19,7 +19,7 @@ numpy = [
    { version = ">=2.1.0", python = "3.13" },
    { version = ">=2.3.0", python = ">=3.14" },
 ]
-onnxruntime-gpu = [
+onnxruntime = [
    { version = ">=1.17.0,!=1.20.0,<1.24", python = "3.10" },
    { version = ">=1.17.0,!=1.20.0,!=1.24.0,!=1.24.1", python = ">=3.11,<3.13" },
    { version = ">1.21.0,!=1.24.0,!=1.24.1", python = "3.13" },
Author	SHA1	Message	Date
George	f61e7ab474	deprecate: drop fallback url to a google bucket (#726 ) * deprecate: drop fallback url to a google bucket * deprecation: add a deprecation warning when a model is loaded using cached gcs data	2026-09-25 02:23:15 +07:00
George	09cab68a47	fix: extend the pool of network errors caught during download (#727 )	2026-09-25 02:14:12 +07:00
George	a4616aef62	fix: fix license for colpali and unicom (#723 )	2026-09-24 16:40:52 +07:00
George	8de28b8f2d	fix: fix mypy (#720 )	2026-09-23 02:44:32 +07:00