From 4875fc8325a431480ed72e7a4a229b56c523888e Mon Sep 17 00:00:00 2001
From: fzowl <zoltan@voyageai.com>
Date: Sun, 16 Aug 2026 21:01:58 +0200
Subject: [PATCH 1/7] feat(voyageai): add contextualized embeddings and refresh
 model support

Route voyage-context-* models to VoyageAI's contextualized embeddings API.
The vectorizer accepts a flat list[str] of chunks and calls contextualized_embed
with enable_auto_chunking=True and chunk_size=32000 so each input document maps
to a single chunk, preserving the one-embedding-per-input embed_many contract.

Also fix multimodal embed_many, which collapsed multiple requested contents into
a single multimodal input; each content is now sent as its own input so the
returned embedding count matches the request count.

Bump the voyageai floor to >=0.5.0 (contextualized auto-chunking) and refresh
docs/examples to current models (rerank-2.5, voyage-context-4).
---
 docs/user_guide/04_vectorizers.ipynb  |  33 +++
 docs/user_guide/06_rerankers.ipynb    |   2 +-
 pyproject.toml                        |   4 +-
 redisvl/utils/vectorize/voyageai.py   |  90 ++++++++-
 tests/integration/test_vectorizers.py | 116 +++++++++++
 uv.lock                               | 278 ++++++++++++--------------
 6 files changed, 365 insertions(+), 158 deletions(-)

diff --git a/docs/user_guide/04_vectorizers.ipynb b/docs/user_guide/04_vectorizers.ipynb
index b87c4c571..5de236b81 100644
--- a/docs/user_guide/04_vectorizers.ipynb
+++ b/docs/user_guide/04_vectorizers.ipynb
@@ -596,6 +596,39 @@
         "print(test[:10])"
       ]
     },
+    {
+      "cell_type": "markdown",
+      "metadata": {},
+      "source": [
+        "#### Contextualized embeddings\n",
+        "\n",
+        "VoyageAI's `voyage-context-*` models produce *contextualized* chunk embeddings: each chunk is embedded with awareness of the other chunks in the same request, which improves retrieval quality for chunked documents. The `VoyageAIVectorizer` automatically routes `voyage-context-*` models to the contextualized embeddings API — just pass your list of chunks to `embed_many` and you get one embedding back per chunk."
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {},
+      "outputs": [],
+      "source": [
+        "# Contextualized embeddings (voyage-context-* models) embed each chunk with\n",
+        "# awareness of the surrounding chunks. Pass a list of chunk strings and the\n",
+        "# vectorizer returns one embedding per chunk. Requires voyageai>=0.5.0.\n",
+        "context_vo = VoyageAIVectorizer(\n",
+        "    model=\"voyage-context-4\",  # See https://docs.voyageai.com/docs/contextualized-chunk-embeddings\n",
+        "    api_config={\"api_key\": api_key},\n",
+        ")\n",
+        "\n",
+        "chunks = [\n",
+        "    \"That is a happy dog\",\n",
+        "    \"That is a happy person\",\n",
+        "    \"Today is a sunny day\",\n",
+        "]\n",
+        "context_embeddings = context_vo.embed_many(chunks, input_type=\"document\")\n",
+        "print(\"Number of embeddings:\", len(context_embeddings))\n",
+        "print(\"Vector dimensions:\", len(context_embeddings[0]))"
+      ]
+    },
     {
       "cell_type": "markdown",
       "metadata": {},
diff --git a/docs/user_guide/06_rerankers.ipynb b/docs/user_guide/06_rerankers.ipynb
index a6f0a7860..982c60c6a 100644
--- a/docs/user_guide/06_rerankers.ipynb
+++ b/docs/user_guide/06_rerankers.ipynb
@@ -378,7 +378,7 @@
    "source": [
     "from redisvl.utils.rerank import VoyageAIReranker\n",
     "\n",
-    "reranker = VoyageAIReranker(model=\"rerank-lite-1\", limit=3, api_config={\"api_key\": api_key})\n",
+    "reranker = VoyageAIReranker(model=\"rerank-2.5\", limit=3, api_config={\"api_key\": api_key})\n",
     "# Please check the available models at https://docs.voyageai.com/docs/reranker"
    ]
   },
diff --git a/pyproject.toml b/pyproject.toml
index 09ff02c9a..bae39cc2b 100644
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -47,7 +47,7 @@ mcp = [
 mistralai = ["mistralai>=1.0.0,<2"]
 openai = ["openai>=1.1.0"]
 cohere = ["cohere>=4.44"]
-voyageai = ["voyageai>=0.2.2"]
+voyageai = ["voyageai>=0.5.0"]
 sentence-transformers = ["sentence-transformers>=3.4.0,<4"]
 langcache = ["langcache>=0.11.0"]
 # Legacy Google backend for the deprecated VertexAIVectorizer. The <2.0.0 cap is
@@ -77,7 +77,7 @@ all = [
     "mistralai>=1.0.0,<2",
     "openai>=1.1.0",
     "cohere>=4.44",
-    "voyageai>=0.2.2",
+    "voyageai>=0.5.0",
     "sentence-transformers>=3.4.0,<4",
     "langcache>=0.11.0",
     "google-cloud-aiplatform>=1.26,<2.0.0",
diff --git a/redisvl/utils/vectorize/voyageai.py b/redisvl/utils/vectorize/voyageai.py
index e50fab7c4..662f211f4 100644
--- a/redisvl/utils/vectorize/voyageai.py
+++ b/redisvl/utils/vectorize/voyageai.py
@@ -1,5 +1,5 @@
 import os
-from typing import TYPE_CHECKING, Any
+from typing import TYPE_CHECKING, Any, cast
 
 from pydantic import ConfigDict
 from tenacity import retry, stop_after_attempt, wait_random_exponential
@@ -50,6 +50,17 @@ class VoyageAIVectorizer(BaseVectorizer):
             input_type="document"
         )
 
+        # Contextualized embeddings (voyage-context-* models) - requires voyageai>=0.5.0
+        # Each input document is embedded with awareness of the others in the batch.
+        context_vectorizer = VoyageAIVectorizer(
+            model="voyage-context-4",
+            api_config={"api_key": "your-voyageai-api-key"}
+        )
+        context_embeddings = context_vectorizer.embed_many(
+            contents=["chunk one", "chunk two", "chunk three"],
+            input_type="document"
+        )
+
         # Multimodal usage - requires Pillow and voyageai>=0.3.6
 
         vectorizer = VoyageAIVectorizer(
@@ -131,6 +142,11 @@ def is_multimodal(self) -> bool:
         """Whether a multimodal model has been configured."""
         return "multimodal" in self.model
 
+    @property
+    def is_context(self) -> bool:
+        """Whether a contextualized-embedding model (voyage-context-*) has been configured."""
+        return "context" in self.model
+
     def embed_image(self, image_path: str, **kwargs) -> list[float] | bytes:
         """Embed an image (from its path on disk) using VoyageAI's multimodal API. Requires pillow to be installed."""
         if not self.is_multimodal:
@@ -343,10 +359,19 @@ def _embed_many(
         try:
             embeddings: list[Any] = []
             for batch in self.batchify(contents, batch_size):
+                if self.is_context:
+                    embeddings.extend(
+                        self._embed_context_batch(batch, input_type, **kwargs)
+                    )
+                    continue
                 response = self._embed_fn(
                     (
-                        [batch] if self.is_multimodal else batch
-                    ),  # Multimodal requires a list of lists/dicts
+                        # Multimodal wraps each item as its own single-part input
+                        # so one embedding is returned per requested content.
+                        [[item] for item in batch]
+                        if self.is_multimodal
+                        else batch
+                    ),
                     model=self.model,
                     input_type=input_type,
                     truncation=truncation,
@@ -359,6 +384,52 @@ def _embed_many(
         except Exception as e:
             raise ValueError(f"Embedding texts failed: {e}")
 
+    def _embed_context_batch(
+        self, batch: list[str], input_type: str | None, **kwargs
+    ) -> list[list[float]]:
+        """Embed a batch with a contextualized (voyage-context-*) model.
+
+        The batch is passed as a flat ``list[str]`` with auto-chunking enabled
+        and a large ``chunk_size`` so each input document resolves to a single
+        chunk, yielding exactly one embedding per requested content.
+        """
+        # contextualized_embed is only present on recent voyageai clients; the
+        # attr-defined ignore keeps mypy happy without vendored stubs.
+        response = self._client.contextualized_embed(  # type: ignore[attr-defined]
+            inputs=batch,
+            model=self.model,
+            input_type=input_type,
+            enable_auto_chunking=True,
+            chunk_size=32000,
+            **kwargs,
+        )
+        # Take the first chunk per document to keep one embedding per input.
+        return cast(
+            "list[list[float]]",
+            [result.embeddings[0] for result in response.results],
+        )
+
+    async def _aembed_context_batch(
+        self, batch: list[str], input_type: str | None, **kwargs
+    ) -> list[list[float]]:
+        """Asynchronously embed a batch with a contextualized model.
+
+        See :meth:`_embed_context_batch` for details on the input format.
+        """
+        response = await self._aclient.contextualized_embed(  # type: ignore[attr-defined]
+            inputs=batch,
+            model=self.model,
+            input_type=input_type,
+            enable_auto_chunking=True,
+            chunk_size=32000,
+            **kwargs,
+        )
+        # Take the first chunk per document to keep one embedding per input.
+        return cast(
+            "list[list[float]]",
+            [result.embeddings[0] for result in response.results],
+        )
+
     async def _aembed(self, content: Any, **kwargs) -> list[float]:
         """
         Asynchronously generate a vector embedding for a single item using the VoyageAI API.
@@ -418,10 +489,19 @@ async def _aembed_many(
         try:
             embeddings: list[Any] = []
             for batch in self.batchify(contents, batch_size):
+                if self.is_context:
+                    embeddings.extend(
+                        await self._aembed_context_batch(batch, input_type, **kwargs)
+                    )
+                    continue
                 response = await self._aembed_fn(
                     (
-                        [batch] if self.is_multimodal else batch
-                    ),  # Multimodal requires a list of lists/dicts
+                        # Multimodal wraps each item as its own single-part input
+                        # so one embedding is returned per requested content.
+                        [[item] for item in batch]
+                        if self.is_multimodal
+                        else batch
+                    ),
                     model=self.model,
                     input_type=input_type,
                     truncation=truncation,
diff --git a/tests/integration/test_vectorizers.py b/tests/integration/test_vectorizers.py
index 599c13378..7f41302a9 100644
--- a/tests/integration/test_vectorizers.py
+++ b/tests/integration/test_vectorizers.py
@@ -659,3 +659,119 @@ def test_deprecated_text_parameter_warning():
         embeddings = vectorizer.embed_many(texts=TEST_TEXTS)
     assert isinstance(embeddings, list)
     assert len(embeddings) == len(TEST_TEXTS)
+
+
+# --- VoyageAI model-routing tests (mocked, no API key required) ---
+
+
+def _fake_voyage_clients(dims=4):
+    """Build mocked sync/async VoyageAI clients returning fixed-size embeddings.
+
+    Each endpoint returns exactly one embedding per requested input so tests can
+    assert that batching never collapses multiple inputs into a single request.
+    """
+    from unittest.mock import AsyncMock, MagicMock
+
+    def _ctx_embed(inputs, model, input_type=None, **kwargs):
+        resp = MagicMock()
+        # contextualized_embed returns one result per input document, each with
+        # a list of per-chunk embeddings.
+        resp.results = [
+            MagicMock(index=i, embeddings=[[0.1] * dims]) for i in range(len(inputs))
+        ]
+        return resp
+
+    def _embed(texts, model=None, input_type=None, truncation=True, **kwargs):
+        resp = MagicMock()
+        resp.embeddings = [[0.1] * dims for _ in texts]
+        return resp
+
+    def _mm_embed(inputs, model, input_type=None, truncation=True, **kwargs):
+        resp = MagicMock()
+        resp.embeddings = [[0.1] * dims for _ in inputs]
+        return resp
+
+    client = MagicMock()
+    client.contextualized_embed.side_effect = _ctx_embed
+    client.embed.side_effect = _embed
+    client.multimodal_embed.side_effect = _mm_embed
+
+    aclient = MagicMock()
+    aclient.contextualized_embed = AsyncMock(side_effect=_ctx_embed)
+    aclient.embed = AsyncMock(side_effect=_embed)
+    aclient.multimodal_embed = AsyncMock(side_effect=_mm_embed)
+
+    return client, aclient
+
+
+def _build_voyage_vectorizer(model):
+    from unittest.mock import patch
+
+    client, aclient = _fake_voyage_clients()
+    with (
+        patch("voyageai.Client", return_value=client),
+        patch("voyageai.AsyncClient", return_value=aclient),
+    ):
+        vectorizer = VoyageAIVectorizer(model=model, api_config={"api_key": "test"})
+    return vectorizer, client, aclient
+
+
+def test_voyageai_context_model_detection():
+    """voyage-context-* models are detected as contextualized models."""
+    ctx_vectorizer, _, _ = _build_voyage_vectorizer("voyage-context-4")
+    assert ctx_vectorizer.is_context is True
+    assert ctx_vectorizer.is_multimodal is False
+
+    plain_vectorizer, _, _ = _build_voyage_vectorizer("voyage-3-large")
+    assert plain_vectorizer.is_context is False
+
+
+def test_voyageai_context_embed_many_uses_contextualized_api():
+    """Context models route to contextualized_embed with auto-chunking enabled."""
+    vectorizer, client, _ = _build_voyage_vectorizer("voyage-context-4")
+
+    embeddings = vectorizer.embed_many(
+        contents=["chunk one", "chunk two", "chunk three"], input_type="document"
+    )
+
+    # One embedding per input, no collapsing.
+    assert len(embeddings) == 3
+
+    _, kwargs = client.contextualized_embed.call_args
+    assert kwargs["inputs"] == ["chunk one", "chunk two", "chunk three"]
+    assert kwargs["enable_auto_chunking"] is True
+    assert kwargs["chunk_size"] == 32000
+    # contextualized_embed does not accept truncation.
+    assert "truncation" not in kwargs
+
+
+@pytest.mark.asyncio
+async def test_voyageai_context_aembed_many_uses_contextualized_api():
+    """Async context models route to contextualized_embed with auto-chunking."""
+    vectorizer, _, aclient = _build_voyage_vectorizer("voyage-context-4")
+
+    embeddings = await vectorizer.aembed_many(
+        contents=["chunk one", "chunk two"], input_type="document"
+    )
+
+    assert len(embeddings) == 2
+    _, kwargs = aclient.contextualized_embed.call_args
+    assert kwargs["inputs"] == ["chunk one", "chunk two"]
+    assert kwargs["enable_auto_chunking"] is True
+    assert kwargs["chunk_size"] == 32000
+
+
+def test_voyageai_multimodal_embed_many_does_not_collapse_inputs():
+    """Multimodal embed_many sends each content as its own input (no collapsing)."""
+    vectorizer, client, _ = _build_voyage_vectorizer("voyage-multimodal-3.5")
+
+    embeddings = vectorizer.embed_many(
+        contents=["Ocean waves", "Forest trees"], input_type="document"
+    )
+
+    # Two requested contents must yield two embeddings.
+    assert len(embeddings) == 2
+
+    args, _ = client.multimodal_embed.call_args
+    # Each item is wrapped as its own single-part multimodal input.
+    assert args[0] == [["Ocean waves"], ["Forest trees"]]
diff --git a/uv.lock b/uv.lock
index 210d06189..02b61e07f 100644
--- a/uv.lock
+++ b/uv.lock
@@ -29,7 +29,7 @@ resolution-markers = [
     "python_full_version < '3.11'",
 ]
 dependencies = [
-    { name = "caio" },
+    { name = "caio", marker = "python_full_version < '3.11'" },
 ]
 sdist = { url = "https://files.pythonhosted.org/packages/67/e2/d7cb819de8df6b5c1968a2756c3cb4122d4fa2b8fc768b53b7c9e5edb646/aiofile-3.9.0.tar.gz", hash = "sha256:e5ad718bb148b265b6df1b3752c4d1d83024b93da9bd599df74b9d9ffcf7919b", size = 17943, upload-time = "2024-10-08T10:39:35.846Z" }
 wheels = [
@@ -47,7 +47,7 @@ resolution-markers = [
     "python_full_version == '3.11.*'",
 ]
 dependencies = [
-    { name = "caio" },
+    { name = "caio", marker = "python_full_version >= '3.11'" },
 ]
 sdist = { url = "https://files.pythonhosted.org/packages/48/41/2fea7e193e061ce54eacc3b7bc0e6a99e4fcff43c78cf0a76dd781ed8334/aiofile-3.11.1.tar.gz", hash = "sha256:1f91912c6643d2a4e49ca4ae3514f0bf3867ce948a36d99a6411b8f4755f4cf9", size = 19342, upload-time = "2026-05-16T08:18:33.538Z" }
 wheels = [
@@ -955,43 +955,43 @@ wheels = [
 
 [package.optional-dependencies]
 cublas = [
-    { name = "nvidia-cublas", marker = "platform_machine == 'aarch64' or platform_machine == 'x86_64'" },
-    { name = "nvidia-cuda-nvrtc", marker = "platform_machine == 'aarch64' or platform_machine == 'x86_64'" },
+    { name = "nvidia-cublas", marker = "(platform_machine == 'aarch64' and sys_platform == 'linux') or (platform_machine == 'x86_64' and sys_platform == 'linux') or (platform_machine == 'AMD64' and sys_platform == 'win32')" },
+    { name = "nvidia-cuda-nvrtc", marker = "(platform_machine == 'aarch64' and sys_platform == 'linux') or (platform_machine == 'x86_64' and sys_platform == 'linux') or (platform_machine == 'AMD64' and sys_platform == 'win32')" },
 ]
 cudart = [
-    { name = "nvidia-cuda-runtime", marker = "platform_machine == 'aarch64' or platform_machine == 'x86_64'" },
+    { name = "nvidia-cuda-runtime", marker = "(platform_machine == 'aarch64' and sys_platform == 'linux') or (platform_machine == 'x86_64' and sys_platform == 'linux') or (platform_machine == 'AMD64' and sys_platform == 'win32')" },
 ]
 cufft = [
-    { name = "nvidia-cufft", marker = "platform_machine == 'aarch64' or platform_machine == 'x86_64'" },
-    { name = "nvidia-nvjitlink", marker = "platform_machine == 'aarch64' or platform_machine == 'x86_64'" },
+    { name = "nvidia-cufft", marker = "(platform_machine == 'aarch64' and sys_platform == 'linux') or (platform_machine == 'x86_64' and sys_platform == 'linux') or (platform_machine == 'AMD64' and sys_platform == 'win32')" },
+    { name = "nvidia-nvjitlink", marker = "(platform_machine == 'aarch64' and sys_platform == 'linux') or (platform_machine == 'x86_64' and sys_platform == 'linux') or (platform_machine == 'AMD64' and sys_platform == 'win32')" },
 ]
 cufile = [
-    { name = "nvidia-cufile", marker = "platform_machine == 'aarch64' or platform_machine == 'x86_64'" },
+    { name = "nvidia-cufile", marker = "(platform_machine == 'aarch64' and sys_platform == 'linux') or (platform_machine == 'x86_64' and sys_platform == 'linux')" },
 ]
 cupti = [
-    { name = "nvidia-cuda-cupti", marker = "platform_machine == 'aarch64' or platform_machine == 'x86_64'" },
+    { name = "nvidia-cuda-cupti", marker = "(platform_machine == 'aarch64' and sys_platform == 'linux') or (platform_machine == 'x86_64' and sys_platform == 'linux') or (platform_machine == 'AMD64' and sys_platform == 'win32')" },
 ]
 curand = [
-    { name = "nvidia-curand", marker = "platform_machine == 'aarch64' or platform_machine == 'x86_64'" },
+    { name = "nvidia-curand", marker = "(platform_machine == 'aarch64' and sys_platform == 'linux') or (platform_machine == 'x86_64' and sys_platform == 'linux') or (platform_machine == 'AMD64' and sys_platform == 'win32')" },
 ]
 cusolver = [
-    { name = "nvidia-cublas", marker = "platform_machine == 'aarch64' or platform_machine == 'x86_64'" },
-    { name = "nvidia-cusolver", marker = "platform_machine == 'aarch64' or platform_machine == 'x86_64'" },
-    { name = "nvidia-cusparse", marker = "platform_machine == 'aarch64' or platform_machine == 'x86_64'" },
-    { name = "nvidia-nvjitlink", marker = "platform_machine == 'aarch64' or platform_machine == 'x86_64'" },
+    { name = "nvidia-cublas", marker = "(platform_machine == 'aarch64' and sys_platform == 'linux') or (platform_machine == 'x86_64' and sys_platform == 'linux') or (platform_machine == 'AMD64' and sys_platform == 'win32')" },
+    { name = "nvidia-cusolver", marker = "(platform_machine == 'aarch64' and sys_platform == 'linux') or (platform_machine == 'x86_64' and sys_platform == 'linux') or (platform_machine == 'AMD64' and sys_platform == 'win32')" },
+    { name = "nvidia-cusparse", marker = "(platform_machine == 'aarch64' and sys_platform == 'linux') or (platform_machine == 'x86_64' and sys_platform == 'linux') or (platform_machine == 'AMD64' and sys_platform == 'win32')" },
+    { name = "nvidia-nvjitlink", marker = "(platform_machine == 'aarch64' and sys_platform == 'linux') or (platform_machine == 'x86_64' and sys_platform == 'linux') or (platform_machine == 'AMD64' and sys_platform == 'win32')" },
 ]
 cusparse = [
-    { name = "nvidia-cusparse", marker = "platform_machine == 'aarch64' or platform_machine == 'x86_64'" },
-    { name = "nvidia-nvjitlink", marker = "platform_machine == 'aarch64' or platform_machine == 'x86_64'" },
+    { name = "nvidia-cusparse", marker = "(platform_machine == 'aarch64' and sys_platform == 'linux') or (platform_machine == 'x86_64' and sys_platform == 'linux') or (platform_machine == 'AMD64' and sys_platform == 'win32')" },
+    { name = "nvidia-nvjitlink", marker = "(platform_machine == 'aarch64' and sys_platform == 'linux') or (platform_machine == 'x86_64' and sys_platform == 'linux') or (platform_machine == 'AMD64' and sys_platform == 'win32')" },
 ]
 nvjitlink = [
-    { name = "nvidia-nvjitlink", marker = "platform_machine == 'aarch64' or platform_machine == 'x86_64'" },
+    { name = "nvidia-nvjitlink", marker = "(platform_machine == 'aarch64' and sys_platform == 'linux') or (platform_machine == 'x86_64' and sys_platform == 'linux') or (platform_machine == 'AMD64' and sys_platform == 'win32')" },
 ]
 nvrtc = [
-    { name = "nvidia-cuda-nvrtc", marker = "platform_machine == 'aarch64' or platform_machine == 'x86_64'" },
+    { name = "nvidia-cuda-nvrtc", marker = "(platform_machine == 'aarch64' and sys_platform == 'linux') or (platform_machine == 'x86_64' and sys_platform == 'linux') or (platform_machine == 'AMD64' and sys_platform == 'win32')" },
 ]
 nvtx = [
-    { name = "nvidia-nvtx", marker = "platform_machine == 'aarch64' or platform_machine == 'x86_64'" },
+    { name = "nvidia-nvtx", marker = "(platform_machine == 'aarch64' and sys_platform == 'linux') or (platform_machine == 'x86_64' and sys_platform == 'linux') or (platform_machine == 'AMD64' and sys_platform == 'win32')" },
 ]
 
 [[package]]
@@ -1299,18 +1299,6 @@ server = [
     { name = "websockets" },
 ]
 
-[[package]]
-name = "ffmpeg-python"
-version = "0.2.0"
-source = { registry = "https://pypi.org/simple" }
-dependencies = [
-    { name = "future" },
-]
-sdist = { url = "https://files.pythonhosted.org/packages/dd/5e/d5f9105d59c1325759d838af4e973695081fbbc97182baf73afc78dec266/ffmpeg-python-0.2.0.tar.gz", hash = "sha256:65225db34627c578ef0e11c8b1eb528bb35e024752f6f10b78c011f6f64c4127", size = 21543, upload-time = "2019-07-06T00:19:08.989Z" }
-wheels = [
-    { url = "https://files.pythonhosted.org/packages/d7/0c/56be52741f75bad4dc6555991fabd2e07b432d333da82c11ad701123888a/ffmpeg_python-0.2.0-py3-none-any.whl", hash = "sha256:ac441a0404e053f8b6a1113a77c0f452f1cfc62f6344a769475ffdc0f56c23c5", size = 25024, upload-time = "2019-07-06T00:19:07.215Z" },
-]
-
 [[package]]
 name = "filelock"
 version = "3.29.0"
@@ -1450,15 +1438,6 @@ wheels = [
     { url = "https://files.pythonhosted.org/packages/d5/0c/043d5e551459da400957a1395e0febbf771446ff34291afcbe3d8be2a279/fsspec-2026.4.0-py3-none-any.whl", hash = "sha256:11ef7bb35dab8a394fde6e608221d5cf3e8499401c249bebaeaad760a1a8dec2", size = 203402, upload-time = "2026-04-29T20:42:36.842Z" },
 ]
 
-[[package]]
-name = "future"
-version = "1.0.0"
-source = { registry = "https://pypi.org/simple" }
-sdist = { url = "https://files.pythonhosted.org/packages/a7/b2/4140c69c6a66432916b26158687e821ba631a4c9273c474343badf84d3ba/future-1.0.0.tar.gz", hash = "sha256:bd2968309307861edae1458a4f8a4f3598c03be43b97521076aebf5d94c07b05", size = 1228490, upload-time = "2024-02-21T11:52:38.461Z" }
-wheels = [
-    { url = "https://files.pythonhosted.org/packages/da/71/ae30dadffc90b9006d77af76b393cb9dfbfc9629f339fc1574a1c52e6806/future-1.0.0-py3-none-any.whl", hash = "sha256:929292d34f5872e70396626ef385ec22355a1fae8ad29e1a734c3e43f9fbc216", size = 491326, upload-time = "2024-02-21T11:52:35.956Z" },
-]
-
 [[package]]
 name = "google-api-core"
 version = "2.25.2"
@@ -1467,11 +1446,11 @@ resolution-markers = [
     "python_full_version >= '3.14'",
 ]
 dependencies = [
-    { name = "google-auth" },
-    { name = "googleapis-common-protos" },
-    { name = "proto-plus" },
-    { name = "protobuf" },
-    { name = "requests" },
+    { name = "google-auth", marker = "python_full_version >= '3.14'" },
+    { name = "googleapis-common-protos", marker = "python_full_version >= '3.14'" },
+    { name = "proto-plus", marker = "python_full_version >= '3.14'" },
+    { name = "protobuf", marker = "python_full_version >= '3.14'" },
+    { name = "requests", marker = "python_full_version >= '3.14'" },
 ]
 sdist = { url = "https://files.pythonhosted.org/packages/09/cd/63f1557235c2440fe0577acdbc32577c5c002684c58c7f4d770a92366a24/google_api_core-2.25.2.tar.gz", hash = "sha256:1c63aa6af0d0d5e37966f157a77f9396d820fba59f9e43e9415bc3dc5baff300", size = 166266, upload-time = "2025-10-03T00:07:34.778Z" }
 wheels = [
@@ -1480,8 +1459,8 @@ wheels = [
 
 [package.optional-dependencies]
 grpc = [
-    { name = "grpcio" },
-    { name = "grpcio-status" },
+    { name = "grpcio", marker = "python_full_version >= '3.14'" },
+    { name = "grpcio-status", marker = "python_full_version >= '3.14'" },
 ]
 
 [[package]]
@@ -1495,11 +1474,11 @@ resolution-markers = [
     "python_full_version < '3.11'",
 ]
 dependencies = [
-    { name = "google-auth" },
-    { name = "googleapis-common-protos" },
-    { name = "proto-plus" },
-    { name = "protobuf" },
-    { name = "requests" },
+    { name = "google-auth", marker = "python_full_version < '3.14'" },
+    { name = "googleapis-common-protos", marker = "python_full_version < '3.14'" },
+    { name = "proto-plus", marker = "python_full_version < '3.14'" },
+    { name = "protobuf", marker = "python_full_version < '3.14'" },
+    { name = "requests", marker = "python_full_version < '3.14'" },
 ]
 sdist = { url = "https://files.pythonhosted.org/packages/16/ce/502a57fb0ec752026d24df1280b162294b22a0afb98a326084f9a979138b/google_api_core-2.30.3.tar.gz", hash = "sha256:e601a37f148585319b26db36e219df68c5d07b6382cff2d580e83404e44d641b", size = 177001, upload-time = "2026-04-10T00:41:28.035Z" }
 wheels = [
@@ -1508,8 +1487,8 @@ wheels = [
 
 [package.optional-dependencies]
 grpc = [
-    { name = "grpcio" },
-    { name = "grpcio-status" },
+    { name = "grpcio", marker = "python_full_version < '3.14'" },
+    { name = "grpcio-status", marker = "python_full_version < '3.14'" },
 ]
 
 [[package]]
@@ -1538,18 +1517,18 @@ resolution-markers = [
     "python_full_version >= '3.14'",
 ]
 dependencies = [
-    { name = "docstring-parser" },
-    { name = "google-api-core", version = "2.25.2", source = { registry = "https://pypi.org/simple" }, extra = ["grpc"] },
-    { name = "google-auth" },
-    { name = "google-cloud-bigquery" },
-    { name = "google-cloud-resource-manager" },
-    { name = "google-cloud-storage", version = "3.4.1", source = { registry = "https://pypi.org/simple" } },
-    { name = "google-genai" },
-    { name = "packaging" },
-    { name = "proto-plus" },
-    { name = "protobuf" },
-    { name = "pydantic" },
-    { name = "typing-extensions" },
+    { name = "docstring-parser", marker = "python_full_version >= '3.14'" },
+    { name = "google-api-core", version = "2.25.2", source = { registry = "https://pypi.org/simple" }, extra = ["grpc"], marker = "python_full_version >= '3.14'" },
+    { name = "google-auth", marker = "python_full_version >= '3.14'" },
+    { name = "google-cloud-bigquery", marker = "python_full_version >= '3.14'" },
+    { name = "google-cloud-resource-manager", marker = "python_full_version >= '3.14'" },
+    { name = "google-cloud-storage", version = "3.4.1", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.14'" },
+    { name = "google-genai", marker = "python_full_version >= '3.14'" },
+    { name = "packaging", marker = "python_full_version >= '3.14'" },
+    { name = "proto-plus", marker = "python_full_version >= '3.14'" },
+    { name = "protobuf", marker = "python_full_version >= '3.14'" },
+    { name = "pydantic", marker = "python_full_version >= '3.14'" },
+    { name = "typing-extensions", marker = "python_full_version >= '3.14'" },
 ]
 sdist = { url = "https://files.pythonhosted.org/packages/c9/f3/b2a9417014c93858a2e3266134f931eefd972c2d410b25d7b8782fc6f143/google_cloud_aiplatform-1.148.1.tar.gz", hash = "sha256:75d605fba34e68714bd08e1e482755d0a6e3ae972805f809d088e686c30879e7", size = 10278758, upload-time = "2026-04-17T23:45:26.738Z" }
 wheels = [
@@ -1567,18 +1546,18 @@ resolution-markers = [
     "python_full_version < '3.11'",
 ]
 dependencies = [
-    { name = "docstring-parser" },
-    { name = "google-api-core", version = "2.30.3", source = { registry = "https://pypi.org/simple" }, extra = ["grpc"] },
-    { name = "google-auth" },
-    { name = "google-cloud-bigquery" },
-    { name = "google-cloud-resource-manager" },
-    { name = "google-cloud-storage", version = "3.10.1", source = { registry = "https://pypi.org/simple" } },
-    { name = "google-genai" },
-    { name = "packaging" },
-    { name = "proto-plus" },
-    { name = "protobuf" },
-    { name = "pydantic" },
-    { name = "typing-extensions" },
+    { name = "docstring-parser", marker = "python_full_version < '3.14'" },
+    { name = "google-api-core", version = "2.30.3", source = { registry = "https://pypi.org/simple" }, extra = ["grpc"], marker = "python_full_version < '3.14'" },
+    { name = "google-auth", marker = "python_full_version < '3.14'" },
+    { name = "google-cloud-bigquery", marker = "python_full_version < '3.14'" },
+    { name = "google-cloud-resource-manager", marker = "python_full_version < '3.14'" },
+    { name = "google-cloud-storage", version = "3.10.1", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.14'" },
+    { name = "google-genai", marker = "python_full_version < '3.14'" },
+    { name = "packaging", marker = "python_full_version < '3.14'" },
+    { name = "proto-plus", marker = "python_full_version < '3.14'" },
+    { name = "protobuf", marker = "python_full_version < '3.14'" },
+    { name = "pydantic", marker = "python_full_version < '3.14'" },
+    { name = "typing-extensions", marker = "python_full_version < '3.14'" },
 ]
 sdist = { url = "https://files.pythonhosted.org/packages/c5/a6/4d5bc1a25069a53383cda1b1da4957765be5a36ad1dce8d726fce557fb10/google_cloud_aiplatform-1.154.0.tar.gz", hash = "sha256:3cfb5afb9006ee202eab93ffc19aeeb111d9e62b574ebf80d7ab91aa9f463677", size = 11020790, upload-time = "2026-05-27T19:20:49.723Z" }
 wheels = [
@@ -1644,12 +1623,12 @@ resolution-markers = [
     "python_full_version >= '3.14'",
 ]
 dependencies = [
-    { name = "google-api-core", version = "2.25.2", source = { registry = "https://pypi.org/simple" } },
-    { name = "google-auth" },
-    { name = "google-cloud-core" },
-    { name = "google-crc32c" },
-    { name = "google-resumable-media" },
-    { name = "requests" },
+    { name = "google-api-core", version = "2.25.2", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.14'" },
+    { name = "google-auth", marker = "python_full_version >= '3.14'" },
+    { name = "google-cloud-core", marker = "python_full_version >= '3.14'" },
+    { name = "google-crc32c", marker = "python_full_version >= '3.14'" },
+    { name = "google-resumable-media", marker = "python_full_version >= '3.14'" },
+    { name = "requests", marker = "python_full_version >= '3.14'" },
 ]
 sdist = { url = "https://files.pythonhosted.org/packages/bd/ef/7cefdca67a6c8b3af0ec38612f9e78e5a9f6179dd91352772ae1a9849246/google_cloud_storage-3.4.1.tar.gz", hash = "sha256:6f041a297e23a4b485fad8c305a7a6e6831855c208bcbe74d00332a909f82268", size = 17238203, upload-time = "2025-10-08T18:43:39.665Z" }
 wheels = [
@@ -1667,12 +1646,12 @@ resolution-markers = [
     "python_full_version < '3.11'",
 ]
 dependencies = [
-    { name = "google-api-core", version = "2.30.3", source = { registry = "https://pypi.org/simple" } },
-    { name = "google-auth" },
-    { name = "google-cloud-core" },
-    { name = "google-crc32c" },
-    { name = "google-resumable-media" },
-    { name = "requests" },
+    { name = "google-api-core", version = "2.30.3", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.14'" },
+    { name = "google-auth", marker = "python_full_version < '3.14'" },
+    { name = "google-cloud-core", marker = "python_full_version < '3.14'" },
+    { name = "google-crc32c", marker = "python_full_version < '3.14'" },
+    { name = "google-resumable-media", marker = "python_full_version < '3.14'" },
+    { name = "requests", marker = "python_full_version < '3.14'" },
 ]
 sdist = { url = "https://files.pythonhosted.org/packages/4c/47/205eb8e9a1739b5345843e5a425775cbdc472cc38e7eda082ba5b8d02450/google_cloud_storage-3.10.1.tar.gz", hash = "sha256:97db9aa4460727982040edd2bd13ff3d5e2260b5331ad22895802da1fc2a5286", size = 17309950, upload-time = "2026-03-23T09:35:23.409Z" }
 wheels = [
@@ -2103,17 +2082,17 @@ resolution-markers = [
     "python_full_version < '3.11'",
 ]
 dependencies = [
-    { name = "colorama", marker = "sys_platform == 'win32'" },
-    { name = "decorator" },
-    { name = "exceptiongroup" },
-    { name = "jedi" },
-    { name = "matplotlib-inline" },
-    { name = "pexpect", marker = "sys_platform != 'emscripten' and sys_platform != 'win32'" },
-    { name = "prompt-toolkit" },
-    { name = "pygments" },
-    { name = "stack-data" },
-    { name = "traitlets" },
-    { name = "typing-extensions" },
+    { name = "colorama", marker = "python_full_version < '3.11' and sys_platform == 'win32'" },
+    { name = "decorator", marker = "python_full_version < '3.11'" },
+    { name = "exceptiongroup", marker = "python_full_version < '3.11'" },
+    { name = "jedi", marker = "python_full_version < '3.11'" },
+    { name = "matplotlib-inline", marker = "python_full_version < '3.11'" },
+    { name = "pexpect", marker = "python_full_version < '3.11' and sys_platform != 'emscripten' and sys_platform != 'win32'" },
+    { name = "prompt-toolkit", marker = "python_full_version < '3.11'" },
+    { name = "pygments", marker = "python_full_version < '3.11'" },
+    { name = "stack-data", marker = "python_full_version < '3.11'" },
+    { name = "traitlets", marker = "python_full_version < '3.11'" },
+    { name = "typing-extensions", marker = "python_full_version < '3.11'" },
 ]
 sdist = { url = "https://files.pythonhosted.org/packages/40/18/f8598d287006885e7136451fdea0755af4ebcbfe342836f24deefaed1164/ipython-8.39.0.tar.gz", hash = "sha256:4110ae96012c379b8b6db898a07e186c40a2a1ef5d57a7fa83166047d9da7624", size = 5513971, upload-time = "2026-03-27T10:02:13.94Z" }
 wheels = [
@@ -2131,18 +2110,18 @@ resolution-markers = [
     "python_full_version == '3.11.*'",
 ]
 dependencies = [
-    { name = "colorama", marker = "sys_platform == 'win32'" },
-    { name = "decorator" },
-    { name = "ipython-pygments-lexers" },
-    { name = "jedi" },
-    { name = "matplotlib-inline" },
-    { name = "pexpect", marker = "sys_platform != 'emscripten' and sys_platform != 'win32'" },
-    { name = "prompt-toolkit" },
-    { name = "psutil" },
-    { name = "pygments" },
-    { name = "stack-data" },
-    { name = "traitlets" },
-    { name = "typing-extensions", marker = "python_full_version < '3.12'" },
+    { name = "colorama", marker = "python_full_version >= '3.11' and sys_platform == 'win32'" },
+    { name = "decorator", marker = "python_full_version >= '3.11'" },
+    { name = "ipython-pygments-lexers", marker = "python_full_version >= '3.11'" },
+    { name = "jedi", marker = "python_full_version >= '3.11'" },
+    { name = "matplotlib-inline", marker = "python_full_version >= '3.11'" },
+    { name = "pexpect", marker = "python_full_version >= '3.11' and sys_platform != 'emscripten' and sys_platform != 'win32'" },
+    { name = "prompt-toolkit", marker = "python_full_version >= '3.11'" },
+    { name = "psutil", marker = "python_full_version >= '3.11'" },
+    { name = "pygments", marker = "python_full_version >= '3.11'" },
+    { name = "stack-data", marker = "python_full_version >= '3.11'" },
+    { name = "traitlets", marker = "python_full_version >= '3.11'" },
+    { name = "typing-extensions", marker = "python_full_version == '3.11.*'" },
 ]
 sdist = { url = "https://files.pythonhosted.org/packages/cd/c4/87cda5842cf5c31837c06ddb588e11c3c35d8ece89b7a0108c06b8c9b00a/ipython-9.13.0.tar.gz", hash = "sha256:7e834b6afc99f020e3f05966ced34792f40267d64cb1ea9043886dab0dde5967", size = 4430549, upload-time = "2026-04-24T12:24:55.221Z" }
 wheels = [
@@ -2154,7 +2133,7 @@ name = "ipython-pygments-lexers"
 version = "1.1.1"
 source = { registry = "https://pypi.org/simple" }
 dependencies = [
-    { name = "pygments" },
+    { name = "pygments", marker = "python_full_version >= '3.11'" },
 ]
 sdist = { url = "https://files.pythonhosted.org/packages/ef/4c/5dd1d8af08107f88c7f741ead7a40854b8ac24ddf9ae850afbcf698aa552/ipython_pygments_lexers-1.1.1.tar.gz", hash = "sha256:09c0138009e56b6854f9535736f4171d855c8c08a563a0dcd8022f78355c7e81", size = 8393, upload-time = "2025-01-17T11:24:34.505Z" }
 wheels = [
@@ -4428,14 +4407,14 @@ resolution-markers = [
     "python_full_version < '3.11'",
 ]
 dependencies = [
-    { name = "accessible-pygments" },
-    { name = "babel" },
-    { name = "beautifulsoup4" },
-    { name = "docutils" },
-    { name = "packaging" },
-    { name = "pygments" },
-    { name = "sphinx" },
-    { name = "typing-extensions" },
+    { name = "accessible-pygments", marker = "python_full_version < '3.11'" },
+    { name = "babel", marker = "python_full_version < '3.11'" },
+    { name = "beautifulsoup4", marker = "python_full_version < '3.11'" },
+    { name = "docutils", marker = "python_full_version < '3.11'" },
+    { name = "packaging", marker = "python_full_version < '3.11'" },
+    { name = "pygments", marker = "python_full_version < '3.11'" },
+    { name = "sphinx", marker = "python_full_version < '3.11'" },
+    { name = "typing-extensions", marker = "python_full_version < '3.11'" },
 ]
 sdist = { url = "https://files.pythonhosted.org/packages/67/ea/3ab478cccacc2e8ef69892c42c44ae547bae089f356c4b47caf61730958d/pydata_sphinx_theme-0.15.4.tar.gz", hash = "sha256:7762ec0ac59df3acecf49fd2f889e1b4565dbce8b88b2e29ee06fdd90645a06d", size = 2400673, upload-time = "2024-06-25T19:28:45.041Z" }
 wheels = [
@@ -4453,13 +4432,13 @@ resolution-markers = [
     "python_full_version == '3.11.*'",
 ]
 dependencies = [
-    { name = "accessible-pygments" },
-    { name = "babel" },
-    { name = "beautifulsoup4" },
-    { name = "docutils" },
-    { name = "pygments" },
-    { name = "sphinx" },
-    { name = "typing-extensions" },
+    { name = "accessible-pygments", marker = "python_full_version >= '3.11'" },
+    { name = "babel", marker = "python_full_version >= '3.11'" },
+    { name = "beautifulsoup4", marker = "python_full_version >= '3.11'" },
+    { name = "docutils", marker = "python_full_version >= '3.11'" },
+    { name = "pygments", marker = "python_full_version >= '3.11'" },
+    { name = "sphinx", marker = "python_full_version >= '3.11'" },
+    { name = "typing-extensions", marker = "python_full_version >= '3.11'" },
 ]
 sdist = { url = "https://files.pythonhosted.org/packages/00/20/bb50f9de3a6de69e6abd6b087b52fa2418a0418b19597601605f855ad044/pydata_sphinx_theme-0.16.1.tar.gz", hash = "sha256:a08b7f0b7f70387219dc659bff0893a7554d5eb39b59d3b8ef37b8401b7642d7", size = 2412693, upload-time = "2024-12-17T10:53:39.537Z" }
 wheels = [
@@ -4987,8 +4966,8 @@ requires-dist = [
     { name = "tenacity", specifier = ">=8.2.2" },
     { name = "urllib3", marker = "extra == 'all'", specifier = "<2.8.0" },
     { name = "urllib3", marker = "extra == 'bedrock'", specifier = "<2.8.0" },
-    { name = "voyageai", marker = "extra == 'all'", specifier = ">=0.2.2" },
-    { name = "voyageai", marker = "extra == 'voyageai'", specifier = ">=0.2.2" },
+    { name = "voyageai", marker = "extra == 'all'", specifier = ">=0.5.0" },
+    { name = "voyageai", marker = "extra == 'voyageai'", specifier = ">=0.5.0" },
 ]
 provides-extras = ["mcp", "mistralai", "openai", "cohere", "voyageai", "sentence-transformers", "langcache", "vertexai", "google-genai", "bedrock", "pillow", "sql-redis", "all", "ollama"]
 
@@ -5379,10 +5358,10 @@ resolution-markers = [
     "python_full_version < '3.11'",
 ]
 dependencies = [
-    { name = "joblib" },
-    { name = "numpy", version = "2.2.6", source = { registry = "https://pypi.org/simple" } },
-    { name = "scipy", version = "1.15.3", source = { registry = "https://pypi.org/simple" } },
-    { name = "threadpoolctl" },
+    { name = "joblib", marker = "python_full_version < '3.11'" },
+    { name = "numpy", version = "2.2.6", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" },
+    { name = "scipy", version = "1.15.3", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" },
+    { name = "threadpoolctl", marker = "python_full_version < '3.11'" },
 ]
 sdist = { url = "https://files.pythonhosted.org/packages/98/c2/a7855e41c9d285dfe86dc50b250978105dce513d6e459ea66a6aeb0e1e0c/scikit_learn-1.7.2.tar.gz", hash = "sha256:20e9e49ecd130598f1ca38a1d85090e1a600147b9c02fa6f15d69cb53d968fda", size = 7193136, upload-time = "2025-09-09T08:21:29.075Z" }
 wheels = [
@@ -5429,10 +5408,10 @@ resolution-markers = [
     "python_full_version == '3.11.*'",
 ]
 dependencies = [
-    { name = "joblib" },
-    { name = "numpy", version = "2.4.6", source = { registry = "https://pypi.org/simple" } },
-    { name = "scipy", version = "1.17.1", source = { registry = "https://pypi.org/simple" } },
-    { name = "threadpoolctl" },
+    { name = "joblib", marker = "python_full_version >= '3.11'" },
+    { name = "numpy", version = "2.4.6", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" },
+    { name = "scipy", version = "1.17.1", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" },
+    { name = "threadpoolctl", marker = "python_full_version >= '3.11'" },
 ]
 sdist = { url = "https://files.pythonhosted.org/packages/0e/d4/40988bf3b8e34feec1d0e6a051446b1f66225f8529b9309becaeef62b6c4/scikit_learn-1.8.0.tar.gz", hash = "sha256:9bccbb3b40e3de10351f8f5068e105d0f4083b1a65fa07b6634fbc401a6287fd", size = 7335585, upload-time = "2025-12-10T07:08:53.618Z" }
 wheels = [
@@ -5482,7 +5461,7 @@ resolution-markers = [
     "python_full_version < '3.11'",
 ]
 dependencies = [
-    { name = "numpy", version = "2.2.6", source = { registry = "https://pypi.org/simple" } },
+    { name = "numpy", version = "2.2.6", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" },
 ]
 sdist = { url = "https://files.pythonhosted.org/packages/0f/37/6964b830433e654ec7485e45a00fc9a27cf868d622838f6b6d9c5ec0d532/scipy-1.15.3.tar.gz", hash = "sha256:eae3cf522bc7df64b42cad3925c876e1b0b6c35c1337c93e12c0f366f55b0eaf", size = 59419214, upload-time = "2025-05-08T16:13:05.955Z" }
 wheels = [
@@ -5544,7 +5523,7 @@ resolution-markers = [
     "python_full_version == '3.11.*'",
 ]
 dependencies = [
-    { name = "numpy", version = "2.4.6", source = { registry = "https://pypi.org/simple" } },
+    { name = "numpy", version = "2.4.6", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" },
 ]
 sdist = { url = "https://files.pythonhosted.org/packages/7a/97/5a3609c4f8d58b039179648e62dd220f89864f56f7357f5d4f45c29eb2cc/scipy-1.17.1.tar.gz", hash = "sha256:95d8e012d8cb8816c226aef832200b1d45109ed4464303e997c5b13122b297c0", size = 30573822, upload-time = "2026-02-23T00:26:24.851Z" }
 wheels = [
@@ -5724,8 +5703,8 @@ resolution-markers = [
     "python_full_version < '3.11'",
 ]
 dependencies = [
-    { name = "pydata-sphinx-theme", version = "0.15.4", source = { registry = "https://pypi.org/simple" } },
-    { name = "sphinx" },
+    { name = "pydata-sphinx-theme", version = "0.15.4", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" },
+    { name = "sphinx", marker = "python_full_version < '3.11'" },
 ]
 sdist = { url = "https://files.pythonhosted.org/packages/45/19/d002ed96bdc7738c15847c730e1e88282d738263deac705d5713b4d8fa94/sphinx_book_theme-1.1.4.tar.gz", hash = "sha256:73efe28af871d0a89bd05856d300e61edce0d5b2fbb7984e84454be0fedfe9ed", size = 439188, upload-time = "2025-02-20T16:32:32.581Z" }
 wheels = [
@@ -5743,8 +5722,8 @@ resolution-markers = [
     "python_full_version == '3.11.*'",
 ]
 dependencies = [
-    { name = "pydata-sphinx-theme", version = "0.16.1", source = { registry = "https://pypi.org/simple" } },
-    { name = "sphinx" },
+    { name = "pydata-sphinx-theme", version = "0.16.1", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" },
+    { name = "sphinx", marker = "python_full_version >= '3.11'" },
 ]
 sdist = { url = "https://files.pythonhosted.org/packages/eb/f7/154786f3cfb7692cd7acc24b6dfe4dcd1146b66f376b17df9e47125555e9/sphinx_book_theme-1.2.0.tar.gz", hash = "sha256:4a7ebfc7da4395309ac942ddfc38fbec5c5254c3be22195e99ad12586fbda9e3", size = 443962, upload-time = "2026-03-09T23:20:30.442Z" }
 wheels = [
@@ -6493,24 +6472,23 @@ wheels = [
 
 [[package]]
 name = "voyageai"
-version = "0.3.7"
+version = "0.5.0"
 source = { registry = "https://pypi.org/simple" }
 dependencies = [
     { name = "aiohttp" },
     { name = "aiolimiter" },
-    { name = "ffmpeg-python" },
     { name = "langchain-text-splitters" },
     { name = "numpy", version = "2.2.6", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" },
-    { name = "numpy", version = "2.4.6", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11' and python_full_version < '3.14'" },
+    { name = "numpy", version = "2.4.6", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" },
     { name = "pillow" },
     { name = "pydantic" },
     { name = "requests" },
     { name = "tenacity" },
     { name = "tokenizers" },
 ]
-sdist = { url = "https://files.pythonhosted.org/packages/94/16/1b46b3cd401e1717a68197c1fe336d7bb4e0a1833f8105e1738f5b1add05/voyageai-0.3.7.tar.gz", hash = "sha256:826cd97f97223f42b5babc5c459c9c80f3a8215ce5c0e007b0b276550f790d24", size = 26485, upload-time = "2025-12-16T18:43:05.26Z" }
+sdist = { url = "https://files.pythonhosted.org/packages/ae/d3/b96ecab961692dd55fc5b54fef916c2bc37a942ef33fcc937779d2aaa161/voyageai-0.5.0.tar.gz", hash = "sha256:ed2775fe9faeb96cc2b3931edc35d76185d19f34f1523d1f70535f0451126ae9", size = 37000, upload-time = "2026-07-10T20:13:14.884Z" }
 wheels = [
-    { url = "https://files.pythonhosted.org/packages/60/64/89f6325666d6836979f94ac88b96fefc7527e02e61abc81359843585e088/voyageai-0.3.7-py3-none-any.whl", hash = "sha256:909f6c033001e5a3b3caf970525bf3614a1bfef9003cf3c3b68207dfdb53e86d", size = 34691, upload-time = "2025-12-16T18:43:04.073Z" },
+    { url = "https://files.pythonhosted.org/packages/0d/c4/732038d97a0ac02d690f295a9ee28f4d0553894478ca6944411428e544f7/voyageai-0.5.0-py3-none-any.whl", hash = "sha256:5d7e8dc3b74cac4646d4a835a65cf8c9764070c90353e322d84a91a16207e78b", size = 46561, upload-time = "2026-07-10T20:13:13.909Z" },
 ]
 
 [[package]]

From 1540de362f6c1fd7bd31fb215106e98e85fae334 Mon Sep 17 00:00:00 2001
From: fzowl <zoltan@voyageai.com>
Date: Sun, 16 Aug 2026 22:10:44 +0200
Subject: [PATCH 2/7] feat(voyageai): recognize voyage-4 embedding models in
 batch sizing

Extend _get_batch_size to route the current voyage-4 family into the same
token-limit tiers as their voyage-3.5 counterparts (voyage-4-lite -> 30,
voyage-4 -> 10); voyage-4-large, voyage-code-4 and voyage-context-* keep the
conservative default. Any VoyageAI model id already works since the vectorizer
passes model strings through; this just refreshes the throughput heuristic and
documents the current catalog on the class docstring.

Add a parametrized test covering the batch-size tiers for current models.
---
 redisvl/utils/vectorize/voyageai.py   | 18 +++++++++++++-----
 tests/integration/test_vectorizers.py | 19 +++++++++++++++++++
 2 files changed, 32 insertions(+), 5 deletions(-)

diff --git a/redisvl/utils/vectorize/voyageai.py b/redisvl/utils/vectorize/voyageai.py
index 662f211f4..9ad84f2f4 100644
--- a/redisvl/utils/vectorize/voyageai.py
+++ b/redisvl/utils/vectorize/voyageai.py
@@ -18,8 +18,14 @@ class VoyageAIVectorizer(BaseVectorizer):
     """The VoyageAIVectorizer class utilizes VoyageAI's API to generate
     embeddings for text and multimodal (text / image / video) data.
 
-    This vectorizer is designed to interact with VoyageAI's /embed and /multimodal_embed APIs,
-    requiring an API key for authentication. The key can be provided
+    This vectorizer is designed to interact with VoyageAI's /embed, /multimodal_embed,
+    and /contextualized_embed APIs. Any model identifier accepted by VoyageAI can be
+    passed via ``model`` - for example the general-purpose ``voyage-4-large`` /
+    ``voyage-4`` / ``voyage-4-lite`` family, domain models such as ``voyage-code-4``,
+    contextualized ``voyage-context-*`` models, and multimodal ``voyage-multimodal-*``
+    models. See https://docs.voyageai.com/docs/embeddings for the current catalog.
+
+    It requires an API key for authentication. The key can be provided
     directly in the `api_config` dictionary or through the `VOYAGE_API_KEY`
     environment variable. User must obtain an API key from VoyageAI's website
     (https://dash.voyageai.com/). Additionally, the `voyageai` python
@@ -265,12 +271,14 @@ def _get_batch_size(self) -> int:
         """
         if self.model in ["voyage-2", "voyage-02"]:
             return 72
-        elif self.model in ["voyage-3-lite", "voyage-3.5-lite"]:
+        elif self.model in ["voyage-3-lite", "voyage-3.5-lite", "voyage-4-lite"]:
             return 30
-        elif self.model in ["voyage-3", "voyage-3.5"]:
+        elif self.model in ["voyage-3", "voyage-3.5", "voyage-4"]:
             return 10
         else:
-            return 7  # Default for other models
+            # Default for other models (e.g. voyage-3-large, voyage-4-large,
+            # voyage-code-*, voyage-finance-2, voyage-law-2, voyage-context-*).
+            return 7
 
     def _validate_input(
         self, contents: list[Any], input_type: str | None, truncation: bool | None
diff --git a/tests/integration/test_vectorizers.py b/tests/integration/test_vectorizers.py
index 7f41302a9..2affaf108 100644
--- a/tests/integration/test_vectorizers.py
+++ b/tests/integration/test_vectorizers.py
@@ -775,3 +775,22 @@ def test_voyageai_multimodal_embed_many_does_not_collapse_inputs():
     args, _ = client.multimodal_embed.call_args
     # Each item is wrapped as its own single-part multimodal input.
     assert args[0] == [["Ocean waves"], ["Forest trees"]]
+
+
+@pytest.mark.parametrize(
+    "model, expected_batch_size",
+    [
+        ("voyage-2", 72),
+        ("voyage-4-lite", 30),
+        ("voyage-3.5-lite", 30),
+        ("voyage-4", 10),
+        ("voyage-3.5", 10),
+        ("voyage-4-large", 7),
+        ("voyage-code-4", 7),
+        ("voyage-3-large", 7),
+    ],
+)
+def test_voyageai_batch_size_for_current_models(model, expected_batch_size):
+    """Current-generation models fall into batch-size tiers matching their token limits."""
+    vectorizer, _, _ = _build_voyage_vectorizer(model)
+    assert vectorizer._get_batch_size() == expected_batch_size

From 40ff9631dfbbe0548846177e178a9f663782d887 Mon Sep 17 00:00:00 2001
From: fzowl <zoltan@voyageai.com>
Date: Sun, 16 Aug 2026 22:24:18 +0200
Subject: [PATCH 3/7] fix(voyageai): gate contextualized auto-chunking to
 document inputs

VoyageAI's contextualized_embed only accepts enable_auto_chunking=True
when input_type="document"; the previous code hardcoded it for every
call, so embedding a query against a voyage-context-* model
(input_type="query", the standard RAG retrieval path) was rejected by
the API.

Enable auto-chunking (with chunk_size=32000) only on the document side
and omit it for queries, which are embedded as a flat list[str] as-is.
Add mocked sync/async tests for the query path, and cover voyage-4-nano
and voyage-context-4 in the batch-size tiers to reflect the current
catalog.
---
 redisvl/utils/vectorize/voyageai.py   | 61 +++++++++++++++++----------
 tests/integration/test_vectorizers.py | 32 ++++++++++++++
 2 files changed, 70 insertions(+), 23 deletions(-)

diff --git a/redisvl/utils/vectorize/voyageai.py b/redisvl/utils/vectorize/voyageai.py
index 9ad84f2f4..dd631c735 100644
--- a/redisvl/utils/vectorize/voyageai.py
+++ b/redisvl/utils/vectorize/voyageai.py
@@ -21,9 +21,10 @@ class VoyageAIVectorizer(BaseVectorizer):
     This vectorizer is designed to interact with VoyageAI's /embed, /multimodal_embed,
     and /contextualized_embed APIs. Any model identifier accepted by VoyageAI can be
     passed via ``model`` - for example the general-purpose ``voyage-4-large`` /
-    ``voyage-4`` / ``voyage-4-lite`` family, domain models such as ``voyage-code-4``,
-    contextualized ``voyage-context-*`` models, and multimodal ``voyage-multimodal-*``
-    models. See https://docs.voyageai.com/docs/embeddings for the current catalog.
+    ``voyage-4`` / ``voyage-4-lite`` / ``voyage-4-nano`` family, domain models such
+    as ``voyage-code-4``, contextualized ``voyage-context-4`` / ``voyage-context-3``
+    models, and multimodal ``voyage-multimodal-*`` models.
+    See https://docs.voyageai.com/docs/embeddings for the current catalog.
 
     It requires an API key for authentication. The key can be provided
     directly in the `api_config` dictionary or through the `VOYAGE_API_KEY`
@@ -66,6 +67,12 @@ class VoyageAIVectorizer(BaseVectorizer):
             contents=["chunk one", "chunk two", "chunk three"],
             input_type="document"
         )
+        # Retrieval queries use input_type="query"; auto-chunking is a
+        # document-only feature, so query inputs are embedded as-is.
+        context_query = context_vectorizer.embed(
+            content="your query text here",
+            input_type="query"
+        )
 
         # Multimodal usage - requires Pillow and voyageai>=0.3.6
 
@@ -392,26 +399,39 @@ def _embed_many(
         except Exception as e:
             raise ValueError(f"Embedding texts failed: {e}")
 
+    def _context_embed_kwargs(
+        self, batch: list[str], input_type: str | None, kwargs: dict[str, Any]
+    ) -> dict[str, Any]:
+        """Build the kwargs for a ``contextualized_embed`` call.
+
+        The batch is passed as a flat ``list[str]``. Auto-chunking is only valid
+        for ``input_type="document"`` (VoyageAI rejects it for queries), so it is
+        enabled with a large ``chunk_size`` on the document side - making each
+        input resolve to a single chunk - and left off otherwise. Either way the
+        first chunk per input is kept, yielding one embedding per requested item.
+        """
+        enable_auto_chunking = input_type == "document"
+        call_kwargs: dict[str, Any] = {
+            "inputs": batch,
+            "model": self.model,
+            "input_type": input_type,
+            "enable_auto_chunking": enable_auto_chunking,
+            **kwargs,
+        }
+        if enable_auto_chunking:
+            call_kwargs["chunk_size"] = 32000
+        return call_kwargs
+
     def _embed_context_batch(
         self, batch: list[str], input_type: str | None, **kwargs
     ) -> list[list[float]]:
-        """Embed a batch with a contextualized (voyage-context-*) model.
-
-        The batch is passed as a flat ``list[str]`` with auto-chunking enabled
-        and a large ``chunk_size`` so each input document resolves to a single
-        chunk, yielding exactly one embedding per requested content.
-        """
+        """Embed a batch with a contextualized (voyage-context-*) model."""
         # contextualized_embed is only present on recent voyageai clients; the
         # attr-defined ignore keeps mypy happy without vendored stubs.
         response = self._client.contextualized_embed(  # type: ignore[attr-defined]
-            inputs=batch,
-            model=self.model,
-            input_type=input_type,
-            enable_auto_chunking=True,
-            chunk_size=32000,
-            **kwargs,
+            **self._context_embed_kwargs(batch, input_type, kwargs),
         )
-        # Take the first chunk per document to keep one embedding per input.
+        # Take the first chunk per input to keep one embedding per input.
         return cast(
             "list[list[float]]",
             [result.embeddings[0] for result in response.results],
@@ -425,14 +445,9 @@ async def _aembed_context_batch(
         See :meth:`_embed_context_batch` for details on the input format.
         """
         response = await self._aclient.contextualized_embed(  # type: ignore[attr-defined]
-            inputs=batch,
-            model=self.model,
-            input_type=input_type,
-            enable_auto_chunking=True,
-            chunk_size=32000,
-            **kwargs,
+            **self._context_embed_kwargs(batch, input_type, kwargs),
         )
-        # Take the first chunk per document to keep one embedding per input.
+        # Take the first chunk per input to keep one embedding per input.
         return cast(
             "list[list[float]]",
             [result.embeddings[0] for result in response.results],
diff --git a/tests/integration/test_vectorizers.py b/tests/integration/test_vectorizers.py
index 2affaf108..456f7bc8a 100644
--- a/tests/integration/test_vectorizers.py
+++ b/tests/integration/test_vectorizers.py
@@ -745,6 +745,21 @@ def test_voyageai_context_embed_many_uses_contextualized_api():
     assert "truncation" not in kwargs
 
 
+def test_voyageai_context_query_disables_auto_chunking():
+    """Query inputs must not enable auto-chunking (document-only per VoyageAI)."""
+    vectorizer, client, _ = _build_voyage_vectorizer("voyage-context-4")
+
+    embedding = vectorizer.embed(content="find similar docs", input_type="query")
+
+    assert len(embedding) == 4
+    _, kwargs = client.contextualized_embed.call_args
+    assert kwargs["inputs"] == ["find similar docs"]
+    assert kwargs["input_type"] == "query"
+    assert kwargs["enable_auto_chunking"] is False
+    # chunk_size is only valid alongside auto-chunking, so it must be omitted.
+    assert "chunk_size" not in kwargs
+
+
 @pytest.mark.asyncio
 async def test_voyageai_context_aembed_many_uses_contextualized_api():
     """Async context models route to contextualized_embed with auto-chunking."""
@@ -761,6 +776,21 @@ async def test_voyageai_context_aembed_many_uses_contextualized_api():
     assert kwargs["chunk_size"] == 32000
 
 
+@pytest.mark.asyncio
+async def test_voyageai_context_aquery_disables_auto_chunking():
+    """Async query inputs must not enable auto-chunking."""
+    vectorizer, _, aclient = _build_voyage_vectorizer("voyage-context-4")
+
+    embedding = await vectorizer.aembed(content="find similar docs", input_type="query")
+
+    assert len(embedding) == 4
+    _, kwargs = aclient.contextualized_embed.call_args
+    assert kwargs["inputs"] == ["find similar docs"]
+    assert kwargs["input_type"] == "query"
+    assert kwargs["enable_auto_chunking"] is False
+    assert "chunk_size" not in kwargs
+
+
 def test_voyageai_multimodal_embed_many_does_not_collapse_inputs():
     """Multimodal embed_many sends each content as its own input (no collapsing)."""
     vectorizer, client, _ = _build_voyage_vectorizer("voyage-multimodal-3.5")
@@ -786,7 +816,9 @@ def test_voyageai_multimodal_embed_many_does_not_collapse_inputs():
         ("voyage-4", 10),
         ("voyage-3.5", 10),
         ("voyage-4-large", 7),
+        ("voyage-4-nano", 7),
         ("voyage-code-4", 7),
+        ("voyage-context-4", 7),
         ("voyage-3-large", 7),
     ],
 )

From a2e5942abcd8ad052f74e165c2595006582b1399 Mon Sep 17 00:00:00 2001
From: fzowl <zoltan@voyageai.com>
Date: Sun, 16 Aug 2026 22:38:06 +0200
Subject: [PATCH 4/7] chore(voyageai): minimize uv.lock diff and note
 context-model caveats

Regenerate uv.lock with the repo-pinned uv (0.12.3) so the lock diff is
limited to the voyageai 0.3.7->0.5.0 bump and its dropped transitive deps
(ffmpeg-python/future), instead of unrelated resolver churn.

Document the contextualized-model caveats in the constructor: long
documents auto-chunk but only the first chunk's embedding is kept, and
truncation is not forwarded to contextualized_embed.
---
 redisvl/utils/vectorize/voyageai.py |   5 +
 uv.lock                             | 244 ++++++++++++++--------------
 2 files changed, 127 insertions(+), 122 deletions(-)

diff --git a/redisvl/utils/vectorize/voyageai.py b/redisvl/utils/vectorize/voyageai.py
index dd631c735..9dd551248 100644
--- a/redisvl/utils/vectorize/voyageai.py
+++ b/redisvl/utils/vectorize/voyageai.py
@@ -144,6 +144,11 @@ def __init__(
         Notes:
             - Multimodal models require voyageai>=0.3.6 to be installed for video embeddings, as well as
                 ffmpeg installed on the system. Image embeddings require pillow to be installed.
+            - Contextualized (``voyage-context-*``) models require voyageai>=0.5.0. To keep the
+                one-embedding-per-input contract, a document longer than ``chunk_size`` (32000 tokens)
+                auto-chunks into multiple chunks but only the first chunk's embedding is kept; the rest
+                are dropped. ``truncation`` is not forwarded to the contextualized API (it does not
+                accept it), so it is silently ignored for these models.
 
         """
         super().__init__(model=model, dtype=dtype, cache=cache)
diff --git a/uv.lock b/uv.lock
index 02b61e07f..f1abe419e 100644
--- a/uv.lock
+++ b/uv.lock
@@ -29,7 +29,7 @@ resolution-markers = [
     "python_full_version < '3.11'",
 ]
 dependencies = [
-    { name = "caio", marker = "python_full_version < '3.11'" },
+    { name = "caio" },
 ]
 sdist = { url = "https://files.pythonhosted.org/packages/67/e2/d7cb819de8df6b5c1968a2756c3cb4122d4fa2b8fc768b53b7c9e5edb646/aiofile-3.9.0.tar.gz", hash = "sha256:e5ad718bb148b265b6df1b3752c4d1d83024b93da9bd599df74b9d9ffcf7919b", size = 17943, upload-time = "2024-10-08T10:39:35.846Z" }
 wheels = [
@@ -47,7 +47,7 @@ resolution-markers = [
     "python_full_version == '3.11.*'",
 ]
 dependencies = [
-    { name = "caio", marker = "python_full_version >= '3.11'" },
+    { name = "caio" },
 ]
 sdist = { url = "https://files.pythonhosted.org/packages/48/41/2fea7e193e061ce54eacc3b7bc0e6a99e4fcff43c78cf0a76dd781ed8334/aiofile-3.11.1.tar.gz", hash = "sha256:1f91912c6643d2a4e49ca4ae3514f0bf3867ce948a36d99a6411b8f4755f4cf9", size = 19342, upload-time = "2026-05-16T08:18:33.538Z" }
 wheels = [
@@ -955,43 +955,43 @@ wheels = [
 
 [package.optional-dependencies]
 cublas = [
-    { name = "nvidia-cublas", marker = "(platform_machine == 'aarch64' and sys_platform == 'linux') or (platform_machine == 'x86_64' and sys_platform == 'linux') or (platform_machine == 'AMD64' and sys_platform == 'win32')" },
-    { name = "nvidia-cuda-nvrtc", marker = "(platform_machine == 'aarch64' and sys_platform == 'linux') or (platform_machine == 'x86_64' and sys_platform == 'linux') or (platform_machine == 'AMD64' and sys_platform == 'win32')" },
+    { name = "nvidia-cublas", marker = "platform_machine == 'aarch64' or platform_machine == 'x86_64'" },
+    { name = "nvidia-cuda-nvrtc", marker = "platform_machine == 'aarch64' or platform_machine == 'x86_64'" },
 ]
 cudart = [
-    { name = "nvidia-cuda-runtime", marker = "(platform_machine == 'aarch64' and sys_platform == 'linux') or (platform_machine == 'x86_64' and sys_platform == 'linux') or (platform_machine == 'AMD64' and sys_platform == 'win32')" },
+    { name = "nvidia-cuda-runtime", marker = "platform_machine == 'aarch64' or platform_machine == 'x86_64'" },
 ]
 cufft = [
-    { name = "nvidia-cufft", marker = "(platform_machine == 'aarch64' and sys_platform == 'linux') or (platform_machine == 'x86_64' and sys_platform == 'linux') or (platform_machine == 'AMD64' and sys_platform == 'win32')" },
-    { name = "nvidia-nvjitlink", marker = "(platform_machine == 'aarch64' and sys_platform == 'linux') or (platform_machine == 'x86_64' and sys_platform == 'linux') or (platform_machine == 'AMD64' and sys_platform == 'win32')" },
+    { name = "nvidia-cufft", marker = "platform_machine == 'aarch64' or platform_machine == 'x86_64'" },
+    { name = "nvidia-nvjitlink", marker = "platform_machine == 'aarch64' or platform_machine == 'x86_64'" },
 ]
 cufile = [
-    { name = "nvidia-cufile", marker = "(platform_machine == 'aarch64' and sys_platform == 'linux') or (platform_machine == 'x86_64' and sys_platform == 'linux')" },
+    { name = "nvidia-cufile", marker = "platform_machine == 'aarch64' or platform_machine == 'x86_64'" },
 ]
 cupti = [
-    { name = "nvidia-cuda-cupti", marker = "(platform_machine == 'aarch64' and sys_platform == 'linux') or (platform_machine == 'x86_64' and sys_platform == 'linux') or (platform_machine == 'AMD64' and sys_platform == 'win32')" },
+    { name = "nvidia-cuda-cupti", marker = "platform_machine == 'aarch64' or platform_machine == 'x86_64'" },
 ]
 curand = [
-    { name = "nvidia-curand", marker = "(platform_machine == 'aarch64' and sys_platform == 'linux') or (platform_machine == 'x86_64' and sys_platform == 'linux') or (platform_machine == 'AMD64' and sys_platform == 'win32')" },
+    { name = "nvidia-curand", marker = "platform_machine == 'aarch64' or platform_machine == 'x86_64'" },
 ]
 cusolver = [
-    { name = "nvidia-cublas", marker = "(platform_machine == 'aarch64' and sys_platform == 'linux') or (platform_machine == 'x86_64' and sys_platform == 'linux') or (platform_machine == 'AMD64' and sys_platform == 'win32')" },
-    { name = "nvidia-cusolver", marker = "(platform_machine == 'aarch64' and sys_platform == 'linux') or (platform_machine == 'x86_64' and sys_platform == 'linux') or (platform_machine == 'AMD64' and sys_platform == 'win32')" },
-    { name = "nvidia-cusparse", marker = "(platform_machine == 'aarch64' and sys_platform == 'linux') or (platform_machine == 'x86_64' and sys_platform == 'linux') or (platform_machine == 'AMD64' and sys_platform == 'win32')" },
-    { name = "nvidia-nvjitlink", marker = "(platform_machine == 'aarch64' and sys_platform == 'linux') or (platform_machine == 'x86_64' and sys_platform == 'linux') or (platform_machine == 'AMD64' and sys_platform == 'win32')" },
+    { name = "nvidia-cublas", marker = "platform_machine == 'aarch64' or platform_machine == 'x86_64'" },
+    { name = "nvidia-cusolver", marker = "platform_machine == 'aarch64' or platform_machine == 'x86_64'" },
+    { name = "nvidia-cusparse", marker = "platform_machine == 'aarch64' or platform_machine == 'x86_64'" },
+    { name = "nvidia-nvjitlink", marker = "platform_machine == 'aarch64' or platform_machine == 'x86_64'" },
 ]
 cusparse = [
-    { name = "nvidia-cusparse", marker = "(platform_machine == 'aarch64' and sys_platform == 'linux') or (platform_machine == 'x86_64' and sys_platform == 'linux') or (platform_machine == 'AMD64' and sys_platform == 'win32')" },
-    { name = "nvidia-nvjitlink", marker = "(platform_machine == 'aarch64' and sys_platform == 'linux') or (platform_machine == 'x86_64' and sys_platform == 'linux') or (platform_machine == 'AMD64' and sys_platform == 'win32')" },
+    { name = "nvidia-cusparse", marker = "platform_machine == 'aarch64' or platform_machine == 'x86_64'" },
+    { name = "nvidia-nvjitlink", marker = "platform_machine == 'aarch64' or platform_machine == 'x86_64'" },
 ]
 nvjitlink = [
-    { name = "nvidia-nvjitlink", marker = "(platform_machine == 'aarch64' and sys_platform == 'linux') or (platform_machine == 'x86_64' and sys_platform == 'linux') or (platform_machine == 'AMD64' and sys_platform == 'win32')" },
+    { name = "nvidia-nvjitlink", marker = "platform_machine == 'aarch64' or platform_machine == 'x86_64'" },
 ]
 nvrtc = [
-    { name = "nvidia-cuda-nvrtc", marker = "(platform_machine == 'aarch64' and sys_platform == 'linux') or (platform_machine == 'x86_64' and sys_platform == 'linux') or (platform_machine == 'AMD64' and sys_platform == 'win32')" },
+    { name = "nvidia-cuda-nvrtc", marker = "platform_machine == 'aarch64' or platform_machine == 'x86_64'" },
 ]
 nvtx = [
-    { name = "nvidia-nvtx", marker = "(platform_machine == 'aarch64' and sys_platform == 'linux') or (platform_machine == 'x86_64' and sys_platform == 'linux') or (platform_machine == 'AMD64' and sys_platform == 'win32')" },
+    { name = "nvidia-nvtx", marker = "platform_machine == 'aarch64' or platform_machine == 'x86_64'" },
 ]
 
 [[package]]
@@ -1446,11 +1446,11 @@ resolution-markers = [
     "python_full_version >= '3.14'",
 ]
 dependencies = [
-    { name = "google-auth", marker = "python_full_version >= '3.14'" },
-    { name = "googleapis-common-protos", marker = "python_full_version >= '3.14'" },
-    { name = "proto-plus", marker = "python_full_version >= '3.14'" },
-    { name = "protobuf", marker = "python_full_version >= '3.14'" },
-    { name = "requests", marker = "python_full_version >= '3.14'" },
+    { name = "google-auth" },
+    { name = "googleapis-common-protos" },
+    { name = "proto-plus" },
+    { name = "protobuf" },
+    { name = "requests" },
 ]
 sdist = { url = "https://files.pythonhosted.org/packages/09/cd/63f1557235c2440fe0577acdbc32577c5c002684c58c7f4d770a92366a24/google_api_core-2.25.2.tar.gz", hash = "sha256:1c63aa6af0d0d5e37966f157a77f9396d820fba59f9e43e9415bc3dc5baff300", size = 166266, upload-time = "2025-10-03T00:07:34.778Z" }
 wheels = [
@@ -1459,8 +1459,8 @@ wheels = [
 
 [package.optional-dependencies]
 grpc = [
-    { name = "grpcio", marker = "python_full_version >= '3.14'" },
-    { name = "grpcio-status", marker = "python_full_version >= '3.14'" },
+    { name = "grpcio" },
+    { name = "grpcio-status" },
 ]
 
 [[package]]
@@ -1474,11 +1474,11 @@ resolution-markers = [
     "python_full_version < '3.11'",
 ]
 dependencies = [
-    { name = "google-auth", marker = "python_full_version < '3.14'" },
-    { name = "googleapis-common-protos", marker = "python_full_version < '3.14'" },
-    { name = "proto-plus", marker = "python_full_version < '3.14'" },
-    { name = "protobuf", marker = "python_full_version < '3.14'" },
-    { name = "requests", marker = "python_full_version < '3.14'" },
+    { name = "google-auth" },
+    { name = "googleapis-common-protos" },
+    { name = "proto-plus" },
+    { name = "protobuf" },
+    { name = "requests" },
 ]
 sdist = { url = "https://files.pythonhosted.org/packages/16/ce/502a57fb0ec752026d24df1280b162294b22a0afb98a326084f9a979138b/google_api_core-2.30.3.tar.gz", hash = "sha256:e601a37f148585319b26db36e219df68c5d07b6382cff2d580e83404e44d641b", size = 177001, upload-time = "2026-04-10T00:41:28.035Z" }
 wheels = [
@@ -1487,8 +1487,8 @@ wheels = [
 
 [package.optional-dependencies]
 grpc = [
-    { name = "grpcio", marker = "python_full_version < '3.14'" },
-    { name = "grpcio-status", marker = "python_full_version < '3.14'" },
+    { name = "grpcio" },
+    { name = "grpcio-status" },
 ]
 
 [[package]]
@@ -1517,18 +1517,18 @@ resolution-markers = [
     "python_full_version >= '3.14'",
 ]
 dependencies = [
-    { name = "docstring-parser", marker = "python_full_version >= '3.14'" },
-    { name = "google-api-core", version = "2.25.2", source = { registry = "https://pypi.org/simple" }, extra = ["grpc"], marker = "python_full_version >= '3.14'" },
-    { name = "google-auth", marker = "python_full_version >= '3.14'" },
-    { name = "google-cloud-bigquery", marker = "python_full_version >= '3.14'" },
-    { name = "google-cloud-resource-manager", marker = "python_full_version >= '3.14'" },
-    { name = "google-cloud-storage", version = "3.4.1", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.14'" },
-    { name = "google-genai", marker = "python_full_version >= '3.14'" },
-    { name = "packaging", marker = "python_full_version >= '3.14'" },
-    { name = "proto-plus", marker = "python_full_version >= '3.14'" },
-    { name = "protobuf", marker = "python_full_version >= '3.14'" },
-    { name = "pydantic", marker = "python_full_version >= '3.14'" },
-    { name = "typing-extensions", marker = "python_full_version >= '3.14'" },
+    { name = "docstring-parser" },
+    { name = "google-api-core", version = "2.25.2", source = { registry = "https://pypi.org/simple" }, extra = ["grpc"] },
+    { name = "google-auth" },
+    { name = "google-cloud-bigquery" },
+    { name = "google-cloud-resource-manager" },
+    { name = "google-cloud-storage", version = "3.4.1", source = { registry = "https://pypi.org/simple" } },
+    { name = "google-genai" },
+    { name = "packaging" },
+    { name = "proto-plus" },
+    { name = "protobuf" },
+    { name = "pydantic" },
+    { name = "typing-extensions" },
 ]
 sdist = { url = "https://files.pythonhosted.org/packages/c9/f3/b2a9417014c93858a2e3266134f931eefd972c2d410b25d7b8782fc6f143/google_cloud_aiplatform-1.148.1.tar.gz", hash = "sha256:75d605fba34e68714bd08e1e482755d0a6e3ae972805f809d088e686c30879e7", size = 10278758, upload-time = "2026-04-17T23:45:26.738Z" }
 wheels = [
@@ -1546,18 +1546,18 @@ resolution-markers = [
     "python_full_version < '3.11'",
 ]
 dependencies = [
-    { name = "docstring-parser", marker = "python_full_version < '3.14'" },
-    { name = "google-api-core", version = "2.30.3", source = { registry = "https://pypi.org/simple" }, extra = ["grpc"], marker = "python_full_version < '3.14'" },
-    { name = "google-auth", marker = "python_full_version < '3.14'" },
-    { name = "google-cloud-bigquery", marker = "python_full_version < '3.14'" },
-    { name = "google-cloud-resource-manager", marker = "python_full_version < '3.14'" },
-    { name = "google-cloud-storage", version = "3.10.1", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.14'" },
-    { name = "google-genai", marker = "python_full_version < '3.14'" },
-    { name = "packaging", marker = "python_full_version < '3.14'" },
-    { name = "proto-plus", marker = "python_full_version < '3.14'" },
-    { name = "protobuf", marker = "python_full_version < '3.14'" },
-    { name = "pydantic", marker = "python_full_version < '3.14'" },
-    { name = "typing-extensions", marker = "python_full_version < '3.14'" },
+    { name = "docstring-parser" },
+    { name = "google-api-core", version = "2.30.3", source = { registry = "https://pypi.org/simple" }, extra = ["grpc"] },
+    { name = "google-auth" },
+    { name = "google-cloud-bigquery" },
+    { name = "google-cloud-resource-manager" },
+    { name = "google-cloud-storage", version = "3.10.1", source = { registry = "https://pypi.org/simple" } },
+    { name = "google-genai" },
+    { name = "packaging" },
+    { name = "proto-plus" },
+    { name = "protobuf" },
+    { name = "pydantic" },
+    { name = "typing-extensions" },
 ]
 sdist = { url = "https://files.pythonhosted.org/packages/c5/a6/4d5bc1a25069a53383cda1b1da4957765be5a36ad1dce8d726fce557fb10/google_cloud_aiplatform-1.154.0.tar.gz", hash = "sha256:3cfb5afb9006ee202eab93ffc19aeeb111d9e62b574ebf80d7ab91aa9f463677", size = 11020790, upload-time = "2026-05-27T19:20:49.723Z" }
 wheels = [
@@ -1623,12 +1623,12 @@ resolution-markers = [
     "python_full_version >= '3.14'",
 ]
 dependencies = [
-    { name = "google-api-core", version = "2.25.2", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.14'" },
-    { name = "google-auth", marker = "python_full_version >= '3.14'" },
-    { name = "google-cloud-core", marker = "python_full_version >= '3.14'" },
-    { name = "google-crc32c", marker = "python_full_version >= '3.14'" },
-    { name = "google-resumable-media", marker = "python_full_version >= '3.14'" },
-    { name = "requests", marker = "python_full_version >= '3.14'" },
+    { name = "google-api-core", version = "2.25.2", source = { registry = "https://pypi.org/simple" } },
+    { name = "google-auth" },
+    { name = "google-cloud-core" },
+    { name = "google-crc32c" },
+    { name = "google-resumable-media" },
+    { name = "requests" },
 ]
 sdist = { url = "https://files.pythonhosted.org/packages/bd/ef/7cefdca67a6c8b3af0ec38612f9e78e5a9f6179dd91352772ae1a9849246/google_cloud_storage-3.4.1.tar.gz", hash = "sha256:6f041a297e23a4b485fad8c305a7a6e6831855c208bcbe74d00332a909f82268", size = 17238203, upload-time = "2025-10-08T18:43:39.665Z" }
 wheels = [
@@ -1646,12 +1646,12 @@ resolution-markers = [
     "python_full_version < '3.11'",
 ]
 dependencies = [
-    { name = "google-api-core", version = "2.30.3", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.14'" },
-    { name = "google-auth", marker = "python_full_version < '3.14'" },
-    { name = "google-cloud-core", marker = "python_full_version < '3.14'" },
-    { name = "google-crc32c", marker = "python_full_version < '3.14'" },
-    { name = "google-resumable-media", marker = "python_full_version < '3.14'" },
-    { name = "requests", marker = "python_full_version < '3.14'" },
+    { name = "google-api-core", version = "2.30.3", source = { registry = "https://pypi.org/simple" } },
+    { name = "google-auth" },
+    { name = "google-cloud-core" },
+    { name = "google-crc32c" },
+    { name = "google-resumable-media" },
+    { name = "requests" },
 ]
 sdist = { url = "https://files.pythonhosted.org/packages/4c/47/205eb8e9a1739b5345843e5a425775cbdc472cc38e7eda082ba5b8d02450/google_cloud_storage-3.10.1.tar.gz", hash = "sha256:97db9aa4460727982040edd2bd13ff3d5e2260b5331ad22895802da1fc2a5286", size = 17309950, upload-time = "2026-03-23T09:35:23.409Z" }
 wheels = [
@@ -2082,17 +2082,17 @@ resolution-markers = [
     "python_full_version < '3.11'",
 ]
 dependencies = [
-    { name = "colorama", marker = "python_full_version < '3.11' and sys_platform == 'win32'" },
-    { name = "decorator", marker = "python_full_version < '3.11'" },
-    { name = "exceptiongroup", marker = "python_full_version < '3.11'" },
-    { name = "jedi", marker = "python_full_version < '3.11'" },
-    { name = "matplotlib-inline", marker = "python_full_version < '3.11'" },
-    { name = "pexpect", marker = "python_full_version < '3.11' and sys_platform != 'emscripten' and sys_platform != 'win32'" },
-    { name = "prompt-toolkit", marker = "python_full_version < '3.11'" },
-    { name = "pygments", marker = "python_full_version < '3.11'" },
-    { name = "stack-data", marker = "python_full_version < '3.11'" },
-    { name = "traitlets", marker = "python_full_version < '3.11'" },
-    { name = "typing-extensions", marker = "python_full_version < '3.11'" },
+    { name = "colorama", marker = "sys_platform == 'win32'" },
+    { name = "decorator" },
+    { name = "exceptiongroup" },
+    { name = "jedi" },
+    { name = "matplotlib-inline" },
+    { name = "pexpect", marker = "sys_platform != 'emscripten' and sys_platform != 'win32'" },
+    { name = "prompt-toolkit" },
+    { name = "pygments" },
+    { name = "stack-data" },
+    { name = "traitlets" },
+    { name = "typing-extensions" },
 ]
 sdist = { url = "https://files.pythonhosted.org/packages/40/18/f8598d287006885e7136451fdea0755af4ebcbfe342836f24deefaed1164/ipython-8.39.0.tar.gz", hash = "sha256:4110ae96012c379b8b6db898a07e186c40a2a1ef5d57a7fa83166047d9da7624", size = 5513971, upload-time = "2026-03-27T10:02:13.94Z" }
 wheels = [
@@ -2110,18 +2110,18 @@ resolution-markers = [
     "python_full_version == '3.11.*'",
 ]
 dependencies = [
-    { name = "colorama", marker = "python_full_version >= '3.11' and sys_platform == 'win32'" },
-    { name = "decorator", marker = "python_full_version >= '3.11'" },
-    { name = "ipython-pygments-lexers", marker = "python_full_version >= '3.11'" },
-    { name = "jedi", marker = "python_full_version >= '3.11'" },
-    { name = "matplotlib-inline", marker = "python_full_version >= '3.11'" },
-    { name = "pexpect", marker = "python_full_version >= '3.11' and sys_platform != 'emscripten' and sys_platform != 'win32'" },
-    { name = "prompt-toolkit", marker = "python_full_version >= '3.11'" },
-    { name = "psutil", marker = "python_full_version >= '3.11'" },
-    { name = "pygments", marker = "python_full_version >= '3.11'" },
-    { name = "stack-data", marker = "python_full_version >= '3.11'" },
-    { name = "traitlets", marker = "python_full_version >= '3.11'" },
-    { name = "typing-extensions", marker = "python_full_version == '3.11.*'" },
+    { name = "colorama", marker = "sys_platform == 'win32'" },
+    { name = "decorator" },
+    { name = "ipython-pygments-lexers" },
+    { name = "jedi" },
+    { name = "matplotlib-inline" },
+    { name = "pexpect", marker = "sys_platform != 'emscripten' and sys_platform != 'win32'" },
+    { name = "prompt-toolkit" },
+    { name = "psutil" },
+    { name = "pygments" },
+    { name = "stack-data" },
+    { name = "traitlets" },
+    { name = "typing-extensions", marker = "python_full_version < '3.12'" },
 ]
 sdist = { url = "https://files.pythonhosted.org/packages/cd/c4/87cda5842cf5c31837c06ddb588e11c3c35d8ece89b7a0108c06b8c9b00a/ipython-9.13.0.tar.gz", hash = "sha256:7e834b6afc99f020e3f05966ced34792f40267d64cb1ea9043886dab0dde5967", size = 4430549, upload-time = "2026-04-24T12:24:55.221Z" }
 wheels = [
@@ -2133,7 +2133,7 @@ name = "ipython-pygments-lexers"
 version = "1.1.1"
 source = { registry = "https://pypi.org/simple" }
 dependencies = [
-    { name = "pygments", marker = "python_full_version >= '3.11'" },
+    { name = "pygments" },
 ]
 sdist = { url = "https://files.pythonhosted.org/packages/ef/4c/5dd1d8af08107f88c7f741ead7a40854b8ac24ddf9ae850afbcf698aa552/ipython_pygments_lexers-1.1.1.tar.gz", hash = "sha256:09c0138009e56b6854f9535736f4171d855c8c08a563a0dcd8022f78355c7e81", size = 8393, upload-time = "2025-01-17T11:24:34.505Z" }
 wheels = [
@@ -4407,14 +4407,14 @@ resolution-markers = [
     "python_full_version < '3.11'",
 ]
 dependencies = [
-    { name = "accessible-pygments", marker = "python_full_version < '3.11'" },
-    { name = "babel", marker = "python_full_version < '3.11'" },
-    { name = "beautifulsoup4", marker = "python_full_version < '3.11'" },
-    { name = "docutils", marker = "python_full_version < '3.11'" },
-    { name = "packaging", marker = "python_full_version < '3.11'" },
-    { name = "pygments", marker = "python_full_version < '3.11'" },
-    { name = "sphinx", marker = "python_full_version < '3.11'" },
-    { name = "typing-extensions", marker = "python_full_version < '3.11'" },
+    { name = "accessible-pygments" },
+    { name = "babel" },
+    { name = "beautifulsoup4" },
+    { name = "docutils" },
+    { name = "packaging" },
+    { name = "pygments" },
+    { name = "sphinx" },
+    { name = "typing-extensions" },
 ]
 sdist = { url = "https://files.pythonhosted.org/packages/67/ea/3ab478cccacc2e8ef69892c42c44ae547bae089f356c4b47caf61730958d/pydata_sphinx_theme-0.15.4.tar.gz", hash = "sha256:7762ec0ac59df3acecf49fd2f889e1b4565dbce8b88b2e29ee06fdd90645a06d", size = 2400673, upload-time = "2024-06-25T19:28:45.041Z" }
 wheels = [
@@ -4432,13 +4432,13 @@ resolution-markers = [
     "python_full_version == '3.11.*'",
 ]
 dependencies = [
-    { name = "accessible-pygments", marker = "python_full_version >= '3.11'" },
-    { name = "babel", marker = "python_full_version >= '3.11'" },
-    { name = "beautifulsoup4", marker = "python_full_version >= '3.11'" },
-    { name = "docutils", marker = "python_full_version >= '3.11'" },
-    { name = "pygments", marker = "python_full_version >= '3.11'" },
-    { name = "sphinx", marker = "python_full_version >= '3.11'" },
-    { name = "typing-extensions", marker = "python_full_version >= '3.11'" },
+    { name = "accessible-pygments" },
+    { name = "babel" },
+    { name = "beautifulsoup4" },
+    { name = "docutils" },
+    { name = "pygments" },
+    { name = "sphinx" },
+    { name = "typing-extensions" },
 ]
 sdist = { url = "https://files.pythonhosted.org/packages/00/20/bb50f9de3a6de69e6abd6b087b52fa2418a0418b19597601605f855ad044/pydata_sphinx_theme-0.16.1.tar.gz", hash = "sha256:a08b7f0b7f70387219dc659bff0893a7554d5eb39b59d3b8ef37b8401b7642d7", size = 2412693, upload-time = "2024-12-17T10:53:39.537Z" }
 wheels = [
@@ -5358,10 +5358,10 @@ resolution-markers = [
     "python_full_version < '3.11'",
 ]
 dependencies = [
-    { name = "joblib", marker = "python_full_version < '3.11'" },
-    { name = "numpy", version = "2.2.6", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" },
-    { name = "scipy", version = "1.15.3", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" },
-    { name = "threadpoolctl", marker = "python_full_version < '3.11'" },
+    { name = "joblib" },
+    { name = "numpy", version = "2.2.6", source = { registry = "https://pypi.org/simple" } },
+    { name = "scipy", version = "1.15.3", source = { registry = "https://pypi.org/simple" } },
+    { name = "threadpoolctl" },
 ]
 sdist = { url = "https://files.pythonhosted.org/packages/98/c2/a7855e41c9d285dfe86dc50b250978105dce513d6e459ea66a6aeb0e1e0c/scikit_learn-1.7.2.tar.gz", hash = "sha256:20e9e49ecd130598f1ca38a1d85090e1a600147b9c02fa6f15d69cb53d968fda", size = 7193136, upload-time = "2025-09-09T08:21:29.075Z" }
 wheels = [
@@ -5408,10 +5408,10 @@ resolution-markers = [
     "python_full_version == '3.11.*'",
 ]
 dependencies = [
-    { name = "joblib", marker = "python_full_version >= '3.11'" },
-    { name = "numpy", version = "2.4.6", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" },
-    { name = "scipy", version = "1.17.1", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" },
-    { name = "threadpoolctl", marker = "python_full_version >= '3.11'" },
+    { name = "joblib" },
+    { name = "numpy", version = "2.4.6", source = { registry = "https://pypi.org/simple" } },
+    { name = "scipy", version = "1.17.1", source = { registry = "https://pypi.org/simple" } },
+    { name = "threadpoolctl" },
 ]
 sdist = { url = "https://files.pythonhosted.org/packages/0e/d4/40988bf3b8e34feec1d0e6a051446b1f66225f8529b9309becaeef62b6c4/scikit_learn-1.8.0.tar.gz", hash = "sha256:9bccbb3b40e3de10351f8f5068e105d0f4083b1a65fa07b6634fbc401a6287fd", size = 7335585, upload-time = "2025-12-10T07:08:53.618Z" }
 wheels = [
@@ -5461,7 +5461,7 @@ resolution-markers = [
     "python_full_version < '3.11'",
 ]
 dependencies = [
-    { name = "numpy", version = "2.2.6", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" },
+    { name = "numpy", version = "2.2.6", source = { registry = "https://pypi.org/simple" } },
 ]
 sdist = { url = "https://files.pythonhosted.org/packages/0f/37/6964b830433e654ec7485e45a00fc9a27cf868d622838f6b6d9c5ec0d532/scipy-1.15.3.tar.gz", hash = "sha256:eae3cf522bc7df64b42cad3925c876e1b0b6c35c1337c93e12c0f366f55b0eaf", size = 59419214, upload-time = "2025-05-08T16:13:05.955Z" }
 wheels = [
@@ -5523,7 +5523,7 @@ resolution-markers = [
     "python_full_version == '3.11.*'",
 ]
 dependencies = [
-    { name = "numpy", version = "2.4.6", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" },
+    { name = "numpy", version = "2.4.6", source = { registry = "https://pypi.org/simple" } },
 ]
 sdist = { url = "https://files.pythonhosted.org/packages/7a/97/5a3609c4f8d58b039179648e62dd220f89864f56f7357f5d4f45c29eb2cc/scipy-1.17.1.tar.gz", hash = "sha256:95d8e012d8cb8816c226aef832200b1d45109ed4464303e997c5b13122b297c0", size = 30573822, upload-time = "2026-02-23T00:26:24.851Z" }
 wheels = [
@@ -5703,8 +5703,8 @@ resolution-markers = [
     "python_full_version < '3.11'",
 ]
 dependencies = [
-    { name = "pydata-sphinx-theme", version = "0.15.4", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" },
-    { name = "sphinx", marker = "python_full_version < '3.11'" },
+    { name = "pydata-sphinx-theme", version = "0.15.4", source = { registry = "https://pypi.org/simple" } },
+    { name = "sphinx" },
 ]
 sdist = { url = "https://files.pythonhosted.org/packages/45/19/d002ed96bdc7738c15847c730e1e88282d738263deac705d5713b4d8fa94/sphinx_book_theme-1.1.4.tar.gz", hash = "sha256:73efe28af871d0a89bd05856d300e61edce0d5b2fbb7984e84454be0fedfe9ed", size = 439188, upload-time = "2025-02-20T16:32:32.581Z" }
 wheels = [
@@ -5722,8 +5722,8 @@ resolution-markers = [
     "python_full_version == '3.11.*'",
 ]
 dependencies = [
-    { name = "pydata-sphinx-theme", version = "0.16.1", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" },
-    { name = "sphinx", marker = "python_full_version >= '3.11'" },
+    { name = "pydata-sphinx-theme", version = "0.16.1", source = { registry = "https://pypi.org/simple" } },
+    { name = "sphinx" },
 ]
 sdist = { url = "https://files.pythonhosted.org/packages/eb/f7/154786f3cfb7692cd7acc24b6dfe4dcd1146b66f376b17df9e47125555e9/sphinx_book_theme-1.2.0.tar.gz", hash = "sha256:4a7ebfc7da4395309ac942ddfc38fbec5c5254c3be22195e99ad12586fbda9e3", size = 443962, upload-time = "2026-03-09T23:20:30.442Z" }
 wheels = [

From 809b7bbc09fd2bd5512a2a65e35e581386bc562a Mon Sep 17 00:00:00 2001
From: fzowl <zoltan@voyageai.com>
Date: Mon, 17 Aug 2026 00:25:53 +0200
Subject: [PATCH 5/7] docs(voyageai): describe contextualized embeddings as
 independent per-input

The notebook claimed each chunk is embedded 'with awareness of the other
chunks in the same request.' That is not what the integration does: every
input string is sent as its own auto-chunked document, so inputs are embedded
independently. Correct the guide to state this plainly, which matches the
one-embedding-per-input contract and cache determinism the vectorizer relies on.
---
 docs/user_guide/04_vectorizers.ipynb | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)

diff --git a/docs/user_guide/04_vectorizers.ipynb b/docs/user_guide/04_vectorizers.ipynb
index 5de236b81..43c672dac 100644
--- a/docs/user_guide/04_vectorizers.ipynb
+++ b/docs/user_guide/04_vectorizers.ipynb
@@ -602,7 +602,7 @@
       "source": [
         "#### Contextualized embeddings\n",
         "\n",
-        "VoyageAI's `voyage-context-*` models produce *contextualized* chunk embeddings: each chunk is embedded with awareness of the other chunks in the same request, which improves retrieval quality for chunked documents. The `VoyageAIVectorizer` automatically routes `voyage-context-*` models to the contextualized embeddings API — just pass your list of chunks to `embed_many` and you get one embedding back per chunk."
+        "VoyageAI's `voyage-context-*` models support *contextualized* chunk embeddings. The `VoyageAIVectorizer` automatically routes `voyage-context-*` models to the contextualized embeddings API: pass your list of chunk strings to `embed_many` and you get one embedding back per chunk. Each input string is sent as its own auto-chunked document, so the inputs are embedded **independently** (a chunk's embedding does not depend on the other strings in the list). This preserves the one-embedding-per-input contract and keeps the embeddings cache deterministic."
       ]
     },
     {
@@ -611,9 +611,9 @@
       "metadata": {},
       "outputs": [],
       "source": [
-        "# Contextualized embeddings (voyage-context-* models) embed each chunk with\n",
-        "# awareness of the surrounding chunks. Pass a list of chunk strings and the\n",
-        "# vectorizer returns one embedding per chunk. Requires voyageai>=0.5.0.\n",
+        "# Contextualized embeddings (voyage-context-* models). Each input string is sent\n",
+        "# as its own auto-chunked document and embedded independently, so the vectorizer\n",
+        "# returns exactly one embedding per input string. Requires voyageai>=0.5.0.\n",
         "context_vo = VoyageAIVectorizer(\n",
         "    model=\"voyage-context-4\",  # See https://docs.voyageai.com/docs/contextualized-chunk-embeddings\n",
         "    api_config={\"api_key\": api_key},\n",

From a9d20eee792e5d57fc5414eb4a45b45cb664edf8 Mon Sep 17 00:00:00 2001
From: fzowl <zoltan@voyageai.com>
Date: Mon, 17 Aug 2026 00:26:00 +0200
Subject: [PATCH 6/7] feat(voyageai): token-aware batching for the text
 embedding path

The plain text embed()/embed_many() path batched only by a fixed per-model
item count, which under-fills small-token batches and can overshoot the
per-request token budget for large inputs. Add token-aware batching that grows
each batch until it would exceed either the item cap or the model's per-request
token limit (VOYAGE_TOKEN_LIMITS, tracking the documented per-model limits),
counting tokens with the VoyageAI tokenizer. A single input larger than the
token limit is still sent alone rather than dropped. Context/multimodal paths
keep item-count batching (auto-chunking / opaque media inputs).

Also clarify the docstrings for contextualized models: inputs are embedded
independently (no cross-input contextualization).

Adds mocked unit tests (run under make test) covering the token boundary split,
the oversized-single-input case, the item-count cap, and sync/async parity.
---
 redisvl/utils/vectorize/voyageai.py   | 104 +++++++++++++++++++++++---
 tests/integration/test_vectorizers.py |  67 +++++++++++++++++
 2 files changed, 161 insertions(+), 10 deletions(-)

diff --git a/redisvl/utils/vectorize/voyageai.py b/redisvl/utils/vectorize/voyageai.py
index 9dd551248..7fcdf50d3 100644
--- a/redisvl/utils/vectorize/voyageai.py
+++ b/redisvl/utils/vectorize/voyageai.py
@@ -13,6 +13,33 @@
 # ignore that voyageai isn't imported
 # mypy: disable-error-code="name-defined"
 
+# Per-request total token limits by model, used for token-aware batching of the
+# plain text embedding path. Values track VoyageAI's documented per-request token
+# limits (https://docs.voyageai.com/docs/embeddings). Unknown models fall back to
+# the conservative default below.
+VOYAGE_TOKEN_LIMITS: dict[str, int] = {
+    "voyage-2": 320_000,
+    "voyage-02": 320_000,
+    "voyage-3": 120_000,
+    "voyage-3-lite": 120_000,
+    "voyage-3-large": 120_000,
+    "voyage-3.5": 320_000,
+    "voyage-3.5-lite": 1_000_000,
+    "voyage-4": 320_000,
+    "voyage-4-lite": 1_000_000,
+    "voyage-4-large": 120_000,
+    "voyage-4-nano": 120_000,
+    "voyage-code-2": 120_000,
+    "voyage-code-3": 120_000,
+    "voyage-code-4": 120_000,
+    "voyage-finance-2": 120_000,
+    "voyage-law-2": 120_000,
+    "voyage-large-2": 120_000,
+    "voyage-large-2-instruct": 120_000,
+    "voyage-multilingual-2": 120_000,
+}
+DEFAULT_VOYAGE_TOKEN_LIMIT = 120_000
+
 
 class VoyageAIVectorizer(BaseVectorizer):
     """The VoyageAIVectorizer class utilizes VoyageAI's API to generate
@@ -58,7 +85,9 @@ class VoyageAIVectorizer(BaseVectorizer):
         )
 
         # Contextualized embeddings (voyage-context-* models) - requires voyageai>=0.5.0
-        # Each input document is embedded with awareness of the others in the batch.
+        # Each input string is treated as its own document (auto-chunked) and
+        # embedded independently: inputs do not influence one another, which keeps
+        # the one-embedding-per-input contract and cache determinism intact.
         context_vectorizer = VoyageAIVectorizer(
             model="voyage-context-4",
             api_config={"api_key": "your-voyageai-api-key"}
@@ -144,11 +173,17 @@ def __init__(
         Notes:
             - Multimodal models require voyageai>=0.3.6 to be installed for video embeddings, as well as
                 ffmpeg installed on the system. Image embeddings require pillow to be installed.
-            - Contextualized (``voyage-context-*``) models require voyageai>=0.5.0. To keep the
-                one-embedding-per-input contract, a document longer than ``chunk_size`` (32000 tokens)
-                auto-chunks into multiple chunks but only the first chunk's embedding is kept; the rest
-                are dropped. ``truncation`` is not forwarded to the contextualized API (it does not
-                accept it), so it is silently ignored for these models.
+            - Contextualized (``voyage-context-*``) models require voyageai>=0.5.0. Each input
+                string is sent as its own document with auto-chunking, so inputs are embedded
+                independently (no cross-input contextualization) and the one-embedding-per-input
+                contract and cache determinism are preserved. A document longer than ``chunk_size``
+                (32000 tokens) auto-chunks into multiple chunks but only the first chunk's embedding
+                is kept; the rest are dropped. ``truncation`` is not forwarded to the contextualized
+                API (it does not accept it), so it is silently ignored for these models.
+            - The plain text embedding path (``embed``/``embed_many`` for non-context,
+                non-multimodal models) uses token-aware batching: inputs are grouped into requests
+                bounded by both the per-model item cap and the model's per-request token limit, so
+                large inputs are packed efficiently without exceeding VoyageAI's token budget.
 
         """
         super().__init__(model=model, dtype=dtype, cache=cache)
@@ -276,10 +311,14 @@ def _set_model_dims(self) -> int:
 
     def _get_batch_size(self) -> int:
         """
-        Determine the appropriate batch size based on the model being used.
+        Determine the per-request item cap for the current model.
+
+        For the plain text path this is combined with the model's per-request
+        token limit (see :meth:`_batchify_by_tokens`); it is the sole bound for
+        the context/multimodal paths.
 
         Returns:
-            int: Recommended batch size for the current model
+            int: Recommended maximum number of items per request
         """
         if self.model in ["voyage-2", "voyage-02"]:
             return 72
@@ -292,6 +331,37 @@ def _get_batch_size(self) -> int:
             # voyage-code-*, voyage-finance-2, voyage-law-2, voyage-context-*).
             return 7
 
+    def _token_limit(self) -> int:
+        """Per-request token budget for the current model (token-aware batching)."""
+        return VOYAGE_TOKEN_LIMITS.get(self.model, DEFAULT_VOYAGE_TOKEN_LIMIT)
+
+    def _count_tokens(self, text: str) -> int:
+        """Count tokens for a single text using VoyageAI's tokenizer."""
+        return len(self._client.tokenize([text], model=self.model)[0])
+
+    def _batchify_by_tokens(self, texts: list[str], batch_size: int):
+        """Yield batches of texts bounded by both item count and token budget.
+
+        Batches are grown until adding the next text would exceed either the
+        per-model item cap (``batch_size``) or the model's per-request token limit.
+        A single text larger than the token limit is still yielded on its own
+        (never dropped), matching VoyageAI's own client batching behavior.
+        """
+        max_tokens = self._token_limit()
+        batch: list[str] = []
+        batch_tokens = 0
+        for text in texts:
+            n_tokens = self._count_tokens(text)
+            if batch and (
+                len(batch) >= batch_size or batch_tokens + n_tokens > max_tokens
+            ):
+                yield batch
+                batch, batch_tokens = [], 0
+            batch.append(text)
+            batch_tokens += n_tokens
+        if batch:
+            yield batch
+
     def _validate_input(
         self, contents: list[Any], input_type: str | None, truncation: bool | None
     ):
@@ -376,9 +446,16 @@ def _embed_many(
         if batch_size is None:
             batch_size = self._get_batch_size()
 
+        # The plain text path uses token-aware batching; context/multimodal paths
+        # stay on fixed item-count batching (auto-chunking / opaque media inputs).
+        if self.is_context or self.is_multimodal:
+            batches: Any = self.batchify(contents, batch_size)
+        else:
+            batches = self._batchify_by_tokens(contents, batch_size)
+
         try:
             embeddings: list[Any] = []
-            for batch in self.batchify(contents, batch_size):
+            for batch in batches:
                 if self.is_context:
                     embeddings.extend(
                         self._embed_context_batch(batch, input_type, **kwargs)
@@ -514,9 +591,16 @@ async def _aembed_many(
         if batch_size is None:
             batch_size = self._get_batch_size()
 
+        # The plain text path uses token-aware batching; context/multimodal paths
+        # stay on fixed item-count batching (auto-chunking / opaque media inputs).
+        if self.is_context or self.is_multimodal:
+            batches: Any = self.batchify(contents, batch_size)
+        else:
+            batches = self._batchify_by_tokens(contents, batch_size)
+
         try:
             embeddings: list[Any] = []
-            for batch in self.batchify(contents, batch_size):
+            for batch in batches:
                 if self.is_context:
                     embeddings.extend(
                         await self._aembed_context_batch(batch, input_type, **kwargs)
diff --git a/tests/integration/test_vectorizers.py b/tests/integration/test_vectorizers.py
index 456f7bc8a..eaf42241e 100644
--- a/tests/integration/test_vectorizers.py
+++ b/tests/integration/test_vectorizers.py
@@ -691,15 +691,22 @@ def _mm_embed(inputs, model, input_type=None, truncation=True, **kwargs):
         resp.embeddings = [[0.1] * dims for _ in inputs]
         return resp
 
+    def _tokenize(texts, model=None):
+        # One token per whitespace-delimited word, so tests can control token
+        # counts by word count (mirrors voyageai's tokenize return shape).
+        return [text.split() for text in texts]
+
     client = MagicMock()
     client.contextualized_embed.side_effect = _ctx_embed
     client.embed.side_effect = _embed
     client.multimodal_embed.side_effect = _mm_embed
+    client.tokenize.side_effect = _tokenize
 
     aclient = MagicMock()
     aclient.contextualized_embed = AsyncMock(side_effect=_ctx_embed)
     aclient.embed = AsyncMock(side_effect=_embed)
     aclient.multimodal_embed = AsyncMock(side_effect=_mm_embed)
+    aclient.tokenize.side_effect = _tokenize
 
     return client, aclient
 
@@ -807,6 +814,66 @@ def test_voyageai_multimodal_embed_many_does_not_collapse_inputs():
     assert args[0] == [["Ocean waves"], ["Forest trees"]]
 
 
+# --- VoyageAI token-aware batching tests (mocked, no API key required) ---
+
+
+def test_voyageai_token_aware_batching_splits_on_token_limit():
+    """Plain text batches split when the per-request token budget is reached."""
+    vectorizer, client, _ = _build_voyage_vectorizer("voyage-3-large")
+    # 3 tokens per text; a 7-token budget fits two texts (6) but not three (9).
+    vectorizer._token_limit = lambda: 7
+    client.embed.reset_mock()
+
+    texts = ["a a a", "b b b", "c c c"]
+    embeddings = vectorizer.embed_many(contents=texts, input_type="document")
+
+    assert len(embeddings) == 3
+    batches = [call.args[0] for call in client.embed.call_args_list]
+    assert batches == [["a a a", "b b b"], ["c c c"]]
+
+
+def test_voyageai_token_aware_batching_oversized_text_goes_alone():
+    """A single text over the token budget is still sent alone, not dropped."""
+    vectorizer, client, _ = _build_voyage_vectorizer("voyage-3-large")
+    vectorizer._token_limit = lambda: 5
+    client.embed.reset_mock()
+
+    texts = ["a a a a a a a a", "b b"]  # 8 tokens (> budget), then 2 tokens
+    embeddings = vectorizer.embed_many(contents=texts, input_type="document")
+
+    assert len(embeddings) == 2
+    batches = [call.args[0] for call in client.embed.call_args_list]
+    assert batches == [["a a a a a a a a"], ["b b"]]
+
+
+def test_voyageai_token_aware_batching_respects_item_cap():
+    """The per-model item cap still bounds a batch when tokens are plentiful."""
+    vectorizer, client, _ = _build_voyage_vectorizer("voyage-4")  # item cap = 10
+    vectorizer._token_limit = lambda: 10_000_000  # effectively unbounded
+    client.embed.reset_mock()
+
+    texts = ["x"] * 25  # 1 token each
+    vectorizer.embed_many(contents=texts, input_type="document")
+
+    batches = [call.args[0] for call in client.embed.call_args_list]
+    assert [len(b) for b in batches] == [10, 10, 5]
+
+
+@pytest.mark.asyncio
+async def test_voyageai_token_aware_batching_async_splits_on_token_limit():
+    """Async plain text batches split on the token budget too (sync/async parity)."""
+    vectorizer, _, aclient = _build_voyage_vectorizer("voyage-3-large")
+    vectorizer._token_limit = lambda: 7
+    aclient.embed.reset_mock()
+
+    texts = ["a a a", "b b b", "c c c"]
+    embeddings = await vectorizer.aembed_many(contents=texts, input_type="document")
+
+    assert len(embeddings) == 3
+    batches = [call.args[0] for call in aclient.embed.call_args_list]
+    assert batches == [["a a a", "b b b"], ["c c c"]]
+
+
 @pytest.mark.parametrize(
     "model, expected_batch_size",
     [

From 7024c585c485feffdc81520d0953077edb75ed48 Mon Sep 17 00:00:00 2001
From: fzowl <zoltan@voyageai.com>
Date: Mon, 17 Aug 2026 22:04:49 +0200
Subject: [PATCH 7/7] fix(voyageai): default contextualized input_type to
 document

Contextualized (voyage-context-*) embedding failed on the default path
where input_type is omitted: VoyageAI rejects a flat list[str] with no
input_type, and callers such as SemanticRouter and SemanticCache never
set one. Treat any non-query input (including the omitted default) as a
document so auto-chunking stays enabled with chunk_size=32000, and keep
skipping auto-chunking only for explicit queries.
---
 redisvl/utils/vectorize/voyageai.py   | 18 ++++++++++-----
 tests/integration/test_vectorizers.py | 33 +++++++++++++++++++++++++++
 2 files changed, 45 insertions(+), 6 deletions(-)

diff --git a/redisvl/utils/vectorize/voyageai.py b/redisvl/utils/vectorize/voyageai.py
index 7fcdf50d3..a681b18e6 100644
--- a/redisvl/utils/vectorize/voyageai.py
+++ b/redisvl/utils/vectorize/voyageai.py
@@ -487,16 +487,22 @@ def _context_embed_kwargs(
         """Build the kwargs for a ``contextualized_embed`` call.
 
         The batch is passed as a flat ``list[str]``. Auto-chunking is only valid
-        for ``input_type="document"`` (VoyageAI rejects it for queries), so it is
-        enabled with a large ``chunk_size`` on the document side - making each
-        input resolve to a single chunk - and left off otherwise. Either way the
-        first chunk per input is kept, yielding one embedding per requested item.
+        for ``input_type="document"`` (VoyageAI rejects it for queries), so every
+        non-query input - including the default where ``input_type`` is omitted -
+        is treated as a document and chunked with a large ``chunk_size``, making
+        each input resolve to a single chunk. Explicit queries skip chunking and
+        are sent as a flat ``list[str]`` as-is. Either way the first chunk per
+        input is kept, yielding one embedding per requested item.
         """
-        enable_auto_chunking = input_type == "document"
+        enable_auto_chunking = input_type != "query"
+        # Auto-chunking requires input_type="document"; a flat list[str] with no
+        # type is rejected, so default (None) callers - e.g. SemanticRouter and
+        # SemanticCache, which never set input_type - resolve to "document".
+        effective_input_type = "document" if enable_auto_chunking else input_type
         call_kwargs: dict[str, Any] = {
             "inputs": batch,
             "model": self.model,
-            "input_type": input_type,
+            "input_type": effective_input_type,
             "enable_auto_chunking": enable_auto_chunking,
             **kwargs,
         }
diff --git a/tests/integration/test_vectorizers.py b/tests/integration/test_vectorizers.py
index 8fc0402de..3880747d2 100644
--- a/tests/integration/test_vectorizers.py
+++ b/tests/integration/test_vectorizers.py
@@ -740,6 +740,39 @@ def test_voyageai_context_query_disables_auto_chunking():
     assert "chunk_size" not in kwargs
 
 
+def test_voyageai_context_default_input_type_enables_auto_chunking():
+    """Omitted input_type must default to document + auto-chunking.
+
+    SemanticRouter / SemanticCache call embed(_many) without input_type; a flat
+    list[str] with no type is rejected by VoyageAI, so the default must resolve
+    to a document (auto-chunking on) rather than passing input_type=None.
+    """
+    vectorizer, client, _ = _build_voyage_vectorizer("voyage-context-4")
+
+    embeddings = vectorizer.embed_many(contents=["chunk one", "chunk two"])
+
+    assert len(embeddings) == 2
+    _, kwargs = client.contextualized_embed.call_args
+    assert kwargs["inputs"] == ["chunk one", "chunk two"]
+    assert kwargs["input_type"] == "document"
+    assert kwargs["enable_auto_chunking"] is True
+    assert kwargs["chunk_size"] == 32000
+
+
+@pytest.mark.asyncio
+async def test_voyageai_context_adefault_input_type_enables_auto_chunking():
+    """Async: omitted input_type must default to document + auto-chunking."""
+    vectorizer, _, aclient = _build_voyage_vectorizer("voyage-context-4")
+
+    embeddings = await vectorizer.aembed_many(contents=["chunk one", "chunk two"])
+
+    assert len(embeddings) == 2
+    _, kwargs = aclient.contextualized_embed.call_args
+    assert kwargs["input_type"] == "document"
+    assert kwargs["enable_auto_chunking"] is True
+    assert kwargs["chunk_size"] == 32000
+
+
 @pytest.mark.asyncio
 async def test_voyageai_context_aembed_many_uses_contextualized_api():
     """Async context models route to contextualized_embed with auto-chunking."""