usecortex-ai 0.3.6__py3-none-any.whl → 0.5.0__py3-none-any.whl

usecortex_ai/embeddings/raw_client.py

@@ -8,6 +8,7 @@ from ..core.client_wrapper import AsyncClientWrapper, SyncClientWrapper
 from ..core.http_response import AsyncHttpResponse, HttpResponse
 from ..core.pydantic_utilities import parse_obj_as
 from ..core.request_options import RequestOptions
+from ..core.serialization import convert_and_respect_annotation_metadata
 from ..errors.bad_request_error import BadRequestError
 from ..errors.forbidden_error import ForbiddenError
 from ..errors.internal_server_error import InternalServerError
@@ -15,11 +16,10 @@ from ..errors.not_found_error import NotFoundError
 from ..errors.service_unavailable_error import ServiceUnavailableError
 from ..errors.unauthorized_error import UnauthorizedError
 from ..errors.unprocessable_entity_error import UnprocessableEntityError
-from ..types.actual_error_response import ActualErrorResponse
-from ..types.embeddings_create_collection_data import EmbeddingsCreateCollectionData
-from ..types.embeddings_delete_data import EmbeddingsDeleteData
-from ..types.embeddings_get_data import EmbeddingsGetData
-from ..types.embeddings_search_data import EmbeddingsSearchData
+from ..types.delete_result import DeleteResult
+from ..types.insert_result import InsertResult
+from ..types.raw_embedding_document import RawEmbeddingDocument
+from ..types.raw_embedding_search_result import RawEmbeddingSearchResult
 
 # this is used as the default value for optional parameters
 OMIT = typing.cast(typing.Any, ...)
@@ -29,45 +29,58 @@ class RawEmbeddingsClient:
     def __init__(self, *, client_wrapper: SyncClientWrapper):
         self._client_wrapper = client_wrapper
 
-    def delete(
+    def insert(
         self,
         *,
-        chunk_ids: typing.Sequence[str],
         tenant_id: str,
+        embeddings: typing.Sequence[RawEmbeddingDocument],
         sub_tenant_id: typing.Optional[str] = OMIT,
+        upsert: typing.Optional[bool] = OMIT,
         request_options: typing.Optional[RequestOptions] = None,
-    ) -> HttpResponse[EmbeddingsDeleteData]:
+    ) -> HttpResponse[InsertResult]:
         """
-        Delete embedding chunks by chunk ID.
+        Upload pre-computed embeddings for advanced similarity search.
+
+        This endpoint accepts vector embeddings that you’ve generated externally,
+         allowing you to integrate with custom embedding models or existing vector databases.
+          The embeddings represent chunks of your content as numerical vectors.
 
-        Use this to remove specific chunks from your embeddings index when they are no longer valid or should not appear in results.
+        The system stores these embeddings and makes them available for semantic search and similarity matching.
+         Use this when you want to leverage specialized embedding models or have existing vector representations.
+        When upsert=True, existing embeddings with the same chunk_id will be updated.
 
         Parameters
         ----------
-        chunk_ids : typing.Sequence[str]
-            The chunk IDs of the source you want to delete
-
         tenant_id : str
             Unique identifier for the tenant/organization
 
+        embeddings : typing.Sequence[RawEmbeddingDocument]
+            List of raw embedding documents to insert
+
         sub_tenant_id : typing.Optional[str]
             Optional sub-tenant identifier used to organize data within a tenant. If omitted, the default sub-tenant created during tenant setup will be used.
 
+        upsert : typing.Optional[bool]
+            If True, update existing embeddings; if False, insert only
+
         request_options : typing.Optional[RequestOptions]
             Request-specific configuration.
 
         Returns
         -------
-        HttpResponse[EmbeddingsDeleteData]
+        HttpResponse[InsertResult]
             Successful Response
         """
         _response = self._client_wrapper.httpx_client.request(
-            "embeddings/delete",
-            method="DELETE",
+            "embeddings/insert-raw-embeddings",
+            method="POST",
             json={
-                "chunk_ids": chunk_ids,
                 "tenant_id": tenant_id,
                 "sub_tenant_id": sub_tenant_id,
+                "embeddings": convert_and_respect_annotation_metadata(
+                    object_=embeddings, annotation=typing.Sequence[RawEmbeddingDocument], direction="write"
+                ),
+                "upsert": upsert,
             },
             headers={
                 "content-type": "application/json",
@@ -78,9 +91,9 @@ class RawEmbeddingsClient:
         try:
             if 200 <= _response.status_code < 300:
                 _data = typing.cast(
-                    EmbeddingsDeleteData,
+                    InsertResult,
                     parse_obj_as(
-                        type_=EmbeddingsDeleteData,  # type: ignore
+                        type_=InsertResult,  # type: ignore
                         object_=_response.json(),
                     ),
                 )
@@ -89,9 +102,9 @@ class RawEmbeddingsClient:
                 raise BadRequestError(
                     headers=dict(_response.headers),
                     body=typing.cast(
-                        ActualErrorResponse,
+                        typing.Optional[typing.Any],
                         parse_obj_as(
-                            type_=ActualErrorResponse,  # type: ignore
+                            type_=typing.Optional[typing.Any],  # type: ignore
                             object_=_response.json(),
                         ),
                     ),
@@ -100,9 +113,9 @@ class RawEmbeddingsClient:
                 raise UnauthorizedError(
                     headers=dict(_response.headers),
                     body=typing.cast(
-                        ActualErrorResponse,
+                        typing.Optional[typing.Any],
                         parse_obj_as(
-                            type_=ActualErrorResponse,  # type: ignore
+                            type_=typing.Optional[typing.Any],  # type: ignore
                             object_=_response.json(),
                         ),
                     ),
@@ -111,9 +124,9 @@ class RawEmbeddingsClient:
                 raise ForbiddenError(
                     headers=dict(_response.headers),
                     body=typing.cast(
-                        ActualErrorResponse,
+                        typing.Optional[typing.Any],
                         parse_obj_as(
-                            type_=ActualErrorResponse,  # type: ignore
+                            type_=typing.Optional[typing.Any],  # type: ignore
                             object_=_response.json(),
                         ),
                     ),
@@ -122,9 +135,9 @@ class RawEmbeddingsClient:
                 raise NotFoundError(
                     headers=dict(_response.headers),
                     body=typing.cast(
-                        ActualErrorResponse,
+                        typing.Optional[typing.Any],
                         parse_obj_as(
-                            type_=ActualErrorResponse,  # type: ignore
+                            type_=typing.Optional[typing.Any],  # type: ignore
                             object_=_response.json(),
                         ),
                     ),
@@ -144,9 +157,9 @@ class RawEmbeddingsClient:
                 raise InternalServerError(
                     headers=dict(_response.headers),
                     body=typing.cast(
-                        ActualErrorResponse,
+                        typing.Optional[typing.Any],
                         parse_obj_as(
-                            type_=ActualErrorResponse,  # type: ignore
+                            type_=typing.Optional[typing.Any],  # type: ignore
                             object_=_response.json(),
                         ),
                     ),
@@ -155,9 +168,9 @@ class RawEmbeddingsClient:
                 raise ServiceUnavailableError(
                     headers=dict(_response.headers),
                     body=typing.cast(
-                        ActualErrorResponse,
+                        typing.Optional[typing.Any],
                         parse_obj_as(
-                            type_=ActualErrorResponse,  # type: ignore
+                            type_=typing.Optional[typing.Any],  # type: ignore
                             object_=_response.json(),
                         ),
                     ),
@@ -171,18 +184,19 @@ class RawEmbeddingsClient:
         self,
         *,
         tenant_id: str,
-        embeddings: typing.Optional[typing.Sequence[float]] = OMIT,
-        sub_tenant_id: typing.Optional[str] = OMIT,
-        max_chunks: typing.Optional[int] = OMIT,
+        sub_tenant_id: str,
+        query_embedding: typing.Sequence[float],
+        limit: typing.Optional[int] = OMIT,
+        filter_expr: typing.Optional[str] = OMIT,
+        output_fields: typing.Optional[typing.Sequence[str]] = OMIT,
         request_options: typing.Optional[RequestOptions] = None,
-    ) -> HttpResponse[EmbeddingsSearchData]:
+    ) -> HttpResponse[typing.List[RawEmbeddingSearchResult]]:
         """
         Find similar chunks using an embedding vector.
 
         Use this to retrieve the most similar chunk IDs to a single query embedding.
 
-
-        Expected outcome
+        Expected outcome:
         - You receive the closest chunk IDs with optional similarity scores.
 
         Parameters
@@ -190,30 +204,39 @@ class RawEmbeddingsClient:
         tenant_id : str
             Unique identifier for the tenant/organization
 
-        embeddings : typing.Optional[typing.Sequence[float]]
-            The embedding vector for search
-
-        sub_tenant_id : typing.Optional[str]
+        sub_tenant_id : str
             Optional sub-tenant identifier used to organize data within a tenant. If omitted, the default sub-tenant created during tenant setup will be used.
 
-        max_chunks : typing.Optional[int]
+        query_embedding : typing.Sequence[float]
+            Query embedding vector to search for
+
+        limit : typing.Optional[int]
+            Maximum number of results to return
+
+        filter_expr : typing.Optional[str]
+            Optional Milvus filter expression for additional filtering
+
+        output_fields : typing.Optional[typing.Sequence[str]]
+            Optional list of fields to return in results (default: chunk_id, source_id, metadata)
 
         request_options : typing.Optional[RequestOptions]
             Request-specific configuration.
 
         Returns
         -------
-        HttpResponse[EmbeddingsSearchData]
+        HttpResponse[typing.List[RawEmbeddingSearchResult]]
             Successful Response
         """
         _response = self._client_wrapper.httpx_client.request(
-            "embeddings/search",
+            "embeddings/search-raw-embeddings",
             method="POST",
             json={
                 "tenant_id": tenant_id,
-                "embeddings": embeddings,
                 "sub_tenant_id": sub_tenant_id,
-                "max_chunks": max_chunks,
+                "query_embedding": query_embedding,
+                "limit": limit,
+                "filter_expr": filter_expr,
+                "output_fields": output_fields,
             },
             headers={
                 "content-type": "application/json",
@@ -224,9 +247,9 @@ class RawEmbeddingsClient:
         try:
             if 200 <= _response.status_code < 300:
                 _data = typing.cast(
-                    EmbeddingsSearchData,
+                    typing.List[RawEmbeddingSearchResult],
                     parse_obj_as(
-                        type_=EmbeddingsSearchData,  # type: ignore
+                        type_=typing.List[RawEmbeddingSearchResult],  # type: ignore
                         object_=_response.json(),
                     ),
                 )
@@ -235,9 +258,9 @@ class RawEmbeddingsClient:
                 raise BadRequestError(
                     headers=dict(_response.headers),
                     body=typing.cast(
-                        ActualErrorResponse,
+                        typing.Optional[typing.Any],
                         parse_obj_as(
-                            type_=ActualErrorResponse,  # type: ignore
+                            type_=typing.Optional[typing.Any],  # type: ignore
                             object_=_response.json(),
                         ),
                     ),
@@ -246,9 +269,9 @@ class RawEmbeddingsClient:
                 raise UnauthorizedError(
                     headers=dict(_response.headers),
                     body=typing.cast(
-                        ActualErrorResponse,
+                        typing.Optional[typing.Any],
                         parse_obj_as(
-                            type_=ActualErrorResponse,  # type: ignore
+                            type_=typing.Optional[typing.Any],  # type: ignore
                             object_=_response.json(),
                         ),
                     ),
@@ -257,9 +280,9 @@ class RawEmbeddingsClient:
                 raise ForbiddenError(
                     headers=dict(_response.headers),
                     body=typing.cast(
-                        ActualErrorResponse,
+                        typing.Optional[typing.Any],
                         parse_obj_as(
-                            type_=ActualErrorResponse,  # type: ignore
+                            type_=typing.Optional[typing.Any],  # type: ignore
                             object_=_response.json(),
                         ),
                     ),
@@ -268,9 +291,9 @@ class RawEmbeddingsClient:
                 raise NotFoundError(
                     headers=dict(_response.headers),
                     body=typing.cast(
-                        ActualErrorResponse,
+                        typing.Optional[typing.Any],
                         parse_obj_as(
-                            type_=ActualErrorResponse,  # type: ignore
+                            type_=typing.Optional[typing.Any],  # type: ignore
                             object_=_response.json(),
                         ),
                     ),
@@ -290,9 +313,9 @@ class RawEmbeddingsClient:
                 raise InternalServerError(
                     headers=dict(_response.headers),
                     body=typing.cast(
-                        ActualErrorResponse,
+                        typing.Optional[typing.Any],
                         parse_obj_as(
-                            type_=ActualErrorResponse,  # type: ignore
+                            type_=typing.Optional[typing.Any],  # type: ignore
                             object_=_response.json(),
                         ),
                     ),
@@ -301,9 +324,9 @@ class RawEmbeddingsClient:
                 raise ServiceUnavailableError(
                     headers=dict(_response.headers),
                     body=typing.cast(
-                        ActualErrorResponse,
+                        typing.Optional[typing.Any],
                         parse_obj_as(
-                            type_=ActualErrorResponse,  # type: ignore
+                            type_=typing.Optional[typing.Any],  # type: ignore
                             object_=_response.json(),
                         ),
                     ),
@@ -313,45 +336,59 @@ class RawEmbeddingsClient:
             raise ApiError(status_code=_response.status_code, headers=dict(_response.headers), body=_response.text)
         raise ApiError(status_code=_response.status_code, headers=dict(_response.headers), body=_response_json)
 
-    def get_by_chunk_ids(
+    def filter(
         self,
         *,
-        chunk_ids: typing.Sequence[str],
         tenant_id: str,
-        sub_tenant_id: typing.Optional[str] = OMIT,
+        sub_tenant_id: str,
+        source_id: typing.Optional[str] = OMIT,
+        chunk_ids: typing.Optional[typing.Sequence[str]] = OMIT,
+        output_fields: typing.Optional[typing.Sequence[str]] = OMIT,
+        limit: typing.Optional[int] = OMIT,
         request_options: typing.Optional[RequestOptions] = None,
-    ) -> HttpResponse[EmbeddingsGetData]:
+    ) -> HttpResponse[typing.List[RawEmbeddingSearchResult]]:
         """
         Retrieve embeddings for specific chunk IDs.
-
         Use this when you already know the chunk IDs and need their corresponding embeddings.
 
         Parameters
         ----------
-        chunk_ids : typing.Sequence[str]
-            The chunk IDs of the source you want to get embeddings for
-
         tenant_id : str
             Unique identifier for the tenant/organization
 
-        sub_tenant_id : typing.Optional[str]
+        sub_tenant_id : str
             Optional sub-tenant identifier used to organize data within a tenant. If omitted, the default sub-tenant created during tenant setup will be used.
 
+        source_id : typing.Optional[str]
+            Optional source ID to filter by (mutually exclusive with chunk_ids)
+
+        chunk_ids : typing.Optional[typing.Sequence[str]]
+            Optional list of chunk IDs to filter by (mutually exclusive with source_id)
+
+        output_fields : typing.Optional[typing.Sequence[str]]
+            Optional list of fields to return in results (default: chunk_id, source_id, metadata)
+
+        limit : typing.Optional[int]
+            Maximum number of results to return
+
         request_options : typing.Optional[RequestOptions]
             Request-specific configuration.
 
         Returns
         -------
-        HttpResponse[EmbeddingsGetData]
+        HttpResponse[typing.List[RawEmbeddingSearchResult]]
             Successful Response
         """
         _response = self._client_wrapper.httpx_client.request(
-            "embeddings/retrieve_by_ids",
+            "embeddings/filter-raw-embeddings",
             method="POST",
             json={
-                "chunk_ids": chunk_ids,
                 "tenant_id": tenant_id,
                 "sub_tenant_id": sub_tenant_id,
+                "source_id": source_id,
+                "chunk_ids": chunk_ids,
+                "output_fields": output_fields,
+                "limit": limit,
             },
             headers={
                 "content-type": "application/json",
@@ -362,9 +399,9 @@ class RawEmbeddingsClient:
         try:
             if 200 <= _response.status_code < 300:
                 _data = typing.cast(
-                    EmbeddingsGetData,
+                    typing.List[RawEmbeddingSearchResult],
                     parse_obj_as(
-                        type_=EmbeddingsGetData,  # type: ignore
+                        type_=typing.List[RawEmbeddingSearchResult],  # type: ignore
                         object_=_response.json(),
                     ),
                 )
@@ -373,9 +410,9 @@ class RawEmbeddingsClient:
                 raise BadRequestError(
                     headers=dict(_response.headers),
                     body=typing.cast(
-                        ActualErrorResponse,
+                        typing.Optional[typing.Any],
                         parse_obj_as(
-                            type_=ActualErrorResponse,  # type: ignore
+                            type_=typing.Optional[typing.Any],  # type: ignore
                             object_=_response.json(),
                         ),
                     ),
@@ -384,9 +421,9 @@ class RawEmbeddingsClient:
                 raise UnauthorizedError(
                     headers=dict(_response.headers),
                     body=typing.cast(
-                        ActualErrorResponse,
+                        typing.Optional[typing.Any],
                         parse_obj_as(
-                            type_=ActualErrorResponse,  # type: ignore
+                            type_=typing.Optional[typing.Any],  # type: ignore
                             object_=_response.json(),
                         ),
                     ),
@@ -395,9 +432,9 @@ class RawEmbeddingsClient:
                 raise ForbiddenError(
                     headers=dict(_response.headers),
                     body=typing.cast(
-                        ActualErrorResponse,
+                        typing.Optional[typing.Any],
                         parse_obj_as(
-                            type_=ActualErrorResponse,  # type: ignore
+                            type_=typing.Optional[typing.Any],  # type: ignore
                             object_=_response.json(),
                         ),
                     ),
@@ -406,9 +443,9 @@ class RawEmbeddingsClient:
                 raise NotFoundError(
                     headers=dict(_response.headers),
                     body=typing.cast(
-                        ActualErrorResponse,
+                        typing.Optional[typing.Any],
                         parse_obj_as(
-                            type_=ActualErrorResponse,  # type: ignore
+                            type_=typing.Optional[typing.Any],  # type: ignore
                             object_=_response.json(),
                         ),
                     ),
@@ -428,9 +465,9 @@ class RawEmbeddingsClient:
                 raise InternalServerError(
                     headers=dict(_response.headers),
                     body=typing.cast(
-                        ActualErrorResponse,
+                        typing.Optional[typing.Any],
                         parse_obj_as(
-                            type_=ActualErrorResponse,  # type: ignore
+                            type_=typing.Optional[typing.Any],  # type: ignore
                             object_=_response.json(),
                         ),
                     ),
@@ -439,9 +476,9 @@ class RawEmbeddingsClient:
                 raise ServiceUnavailableError(
                     headers=dict(_response.headers),
                     body=typing.cast(
-                        ActualErrorResponse,
+                        typing.Optional[typing.Any],
                         parse_obj_as(
-                            type_=ActualErrorResponse,  # type: ignore
+                            type_=typing.Optional[typing.Any],  # type: ignore
                             object_=_response.json(),
                         ),
                     ),
@@ -451,41 +488,60 @@ class RawEmbeddingsClient:
             raise ApiError(status_code=_response.status_code, headers=dict(_response.headers), body=_response.text)
         raise ApiError(status_code=_response.status_code, headers=dict(_response.headers), body=_response_json)
 
-    def create_collection(
-        self, *, tenant_id: str, request_options: typing.Optional[RequestOptions] = None
-    ) -> HttpResponse[EmbeddingsCreateCollectionData]:
+    def delete(
+        self,
+        *,
+        tenant_id: str,
+        sub_tenant_id: typing.Optional[str] = None,
+        source_id: typing.Optional[str] = None,
+        chunk_ids: typing.Optional[typing.Union[str, typing.Sequence[str]]] = None,
+        request_options: typing.Optional[RequestOptions] = None,
+    ) -> HttpResponse[DeleteResult]:
         """
-        Create an embeddings collection for a tenant.
+        Delete embedding chunks by chunk ID.
 
-        Use this to initialize storage so you can index and query embeddings for the tenant.
+        Use this to remove specific chunks from your embeddings index when they are no longer
+        valid or should not appear in results.
 
         Parameters
         ----------
         tenant_id : str
             Unique identifier for the tenant/organization
 
+        sub_tenant_id : typing.Optional[str]
+            Optional sub-tenant ID for scoping deletion
+
+        source_id : typing.Optional[str]
+            Optional source ID to delete by (mutually exclusive with chunk_ids)
+
+        chunk_ids : typing.Optional[typing.Union[str, typing.Sequence[str]]]
+            Optional list of chunk IDs to delete (mutually exclusive with source_id)
+
         request_options : typing.Optional[RequestOptions]
             Request-specific configuration.
 
         Returns
         -------
-        HttpResponse[EmbeddingsCreateCollectionData]
+        HttpResponse[DeleteResult]
             Successful Response
         """
         _response = self._client_wrapper.httpx_client.request(
-            "embeddings/create_tenant",
-            method="POST",
+            "embeddings/delete-raw-embeddings",
+            method="DELETE",
             params={
                 "tenant_id": tenant_id,
+                "sub_tenant_id": sub_tenant_id,
+                "source_id": source_id,
+                "chunk_ids": chunk_ids,
             },
             request_options=request_options,
         )
         try:
             if 200 <= _response.status_code < 300:
                 _data = typing.cast(
-                    EmbeddingsCreateCollectionData,
+                    DeleteResult,
                     parse_obj_as(
-                        type_=EmbeddingsCreateCollectionData,  # type: ignore
+                        type_=DeleteResult,  # type: ignore
                         object_=_response.json(),
                     ),
                 )
@@ -494,9 +550,9 @@ class RawEmbeddingsClient:
                 raise BadRequestError(
                     headers=dict(_response.headers),
                     body=typing.cast(
-                        ActualErrorResponse,
+                        typing.Optional[typing.Any],
                         parse_obj_as(
-                            type_=ActualErrorResponse,  # type: ignore
+                            type_=typing.Optional[typing.Any],  # type: ignore
                             object_=_response.json(),
                         ),
                     ),
@@ -505,9 +561,9 @@ class RawEmbeddingsClient:
                 raise UnauthorizedError(
                     headers=dict(_response.headers),
                     body=typing.cast(
-                        ActualErrorResponse,
+                        typing.Optional[typing.Any],
                         parse_obj_as(
-                            type_=ActualErrorResponse,  # type: ignore
+                            type_=typing.Optional[typing.Any],  # type: ignore
                             object_=_response.json(),
                         ),
                     ),
@@ -516,9 +572,9 @@ class RawEmbeddingsClient:
                 raise ForbiddenError(
                     headers=dict(_response.headers),
                     body=typing.cast(
-                        ActualErrorResponse,
+                        typing.Optional[typing.Any],
                         parse_obj_as(
-                            type_=ActualErrorResponse,  # type: ignore
+                            type_=typing.Optional[typing.Any],  # type: ignore
                             object_=_response.json(),
                         ),
                     ),
@@ -527,9 +583,9 @@ class RawEmbeddingsClient:
                 raise NotFoundError(
                     headers=dict(_response.headers),
                     body=typing.cast(
-                        ActualErrorResponse,
+                        typing.Optional[typing.Any],
                         parse_obj_as(
-                            type_=ActualErrorResponse,  # type: ignore
+                            type_=typing.Optional[typing.Any],  # type: ignore
                             object_=_response.json(),
                         ),
                     ),
@@ -549,9 +605,9 @@ class RawEmbeddingsClient:
                 raise InternalServerError(
                     headers=dict(_response.headers),
                     body=typing.cast(
-                        ActualErrorResponse,
+                        typing.Optional[typing.Any],
                         parse_obj_as(
-                            type_=ActualErrorResponse,  # type: ignore
+                            type_=typing.Optional[typing.Any],  # type: ignore
                             object_=_response.json(),
                         ),
                     ),
@@ -560,9 +616,9 @@ class RawEmbeddingsClient:
                 raise ServiceUnavailableError(
                     headers=dict(_response.headers),
                     body=typing.cast(
-                        ActualErrorResponse,
+                        typing.Optional[typing.Any],
                         parse_obj_as(
-                            type_=ActualErrorResponse,  # type: ignore
+                            type_=typing.Optional[typing.Any],  # type: ignore
                             object_=_response.json(),
                         ),
                     ),
@@ -577,45 +633,58 @@ class AsyncRawEmbeddingsClient:
     def __init__(self, *, client_wrapper: AsyncClientWrapper):
         self._client_wrapper = client_wrapper
 
-    async def delete(
+    async def insert(
         self,
         *,
-        chunk_ids: typing.Sequence[str],
         tenant_id: str,
+        embeddings: typing.Sequence[RawEmbeddingDocument],
         sub_tenant_id: typing.Optional[str] = OMIT,
+        upsert: typing.Optional[bool] = OMIT,
         request_options: typing.Optional[RequestOptions] = None,
-    ) -> AsyncHttpResponse[EmbeddingsDeleteData]:
+    ) -> AsyncHttpResponse[InsertResult]:
         """
-        Delete embedding chunks by chunk ID.
+        Upload pre-computed embeddings for advanced similarity search.
+
+        This endpoint accepts vector embeddings that you’ve generated externally,
+         allowing you to integrate with custom embedding models or existing vector databases.
+          The embeddings represent chunks of your content as numerical vectors.
 
-        Use this to remove specific chunks from your embeddings index when they are no longer valid or should not appear in results.
+        The system stores these embeddings and makes them available for semantic search and similarity matching.
+         Use this when you want to leverage specialized embedding models or have existing vector representations.
+        When upsert=True, existing embeddings with the same chunk_id will be updated.
 
         Parameters
         ----------
-        chunk_ids : typing.Sequence[str]
-            The chunk IDs of the source you want to delete
-
         tenant_id : str
             Unique identifier for the tenant/organization
 
+        embeddings : typing.Sequence[RawEmbeddingDocument]
+            List of raw embedding documents to insert
+
         sub_tenant_id : typing.Optional[str]
             Optional sub-tenant identifier used to organize data within a tenant. If omitted, the default sub-tenant created during tenant setup will be used.
 
+        upsert : typing.Optional[bool]
+            If True, update existing embeddings; if False, insert only
+
         request_options : typing.Optional[RequestOptions]
             Request-specific configuration.
 
         Returns
         -------
-        AsyncHttpResponse[EmbeddingsDeleteData]
+        AsyncHttpResponse[InsertResult]
             Successful Response
         """
         _response = await self._client_wrapper.httpx_client.request(
-            "embeddings/delete",
-            method="DELETE",
+            "embeddings/insert-raw-embeddings",
+            method="POST",
             json={
-                "chunk_ids": chunk_ids,
                 "tenant_id": tenant_id,
                 "sub_tenant_id": sub_tenant_id,
+                "embeddings": convert_and_respect_annotation_metadata(
+                    object_=embeddings, annotation=typing.Sequence[RawEmbeddingDocument], direction="write"
+                ),
+                "upsert": upsert,
             },
             headers={
                 "content-type": "application/json",
@@ -626,9 +695,9 @@ class AsyncRawEmbeddingsClient:
         try:
             if 200 <= _response.status_code < 300:
                 _data = typing.cast(
-                    EmbeddingsDeleteData,
+                    InsertResult,
                     parse_obj_as(
-                        type_=EmbeddingsDeleteData,  # type: ignore
+                        type_=InsertResult,  # type: ignore
                         object_=_response.json(),
                     ),
                 )
@@ -637,9 +706,9 @@ class AsyncRawEmbeddingsClient:
                 raise BadRequestError(
                     headers=dict(_response.headers),
                     body=typing.cast(
-                        ActualErrorResponse,
+                        typing.Optional[typing.Any],
                         parse_obj_as(
-                            type_=ActualErrorResponse,  # type: ignore
+                            type_=typing.Optional[typing.Any],  # type: ignore
                             object_=_response.json(),
                         ),
                     ),
@@ -648,9 +717,9 @@ class AsyncRawEmbeddingsClient:
                 raise UnauthorizedError(
                     headers=dict(_response.headers),
                     body=typing.cast(
-                        ActualErrorResponse,
+                        typing.Optional[typing.Any],
                         parse_obj_as(
-                            type_=ActualErrorResponse,  # type: ignore
+                            type_=typing.Optional[typing.Any],  # type: ignore
                             object_=_response.json(),
                         ),
                     ),
@@ -659,9 +728,9 @@ class AsyncRawEmbeddingsClient:
                 raise ForbiddenError(
                     headers=dict(_response.headers),
                     body=typing.cast(
-                        ActualErrorResponse,
+                        typing.Optional[typing.Any],
                         parse_obj_as(
-                            type_=ActualErrorResponse,  # type: ignore
+                            type_=typing.Optional[typing.Any],  # type: ignore
                             object_=_response.json(),
                         ),
                     ),
@@ -670,9 +739,9 @@ class AsyncRawEmbeddingsClient:
                 raise NotFoundError(
                     headers=dict(_response.headers),
                     body=typing.cast(
-                        ActualErrorResponse,
+                        typing.Optional[typing.Any],
                         parse_obj_as(
-                            type_=ActualErrorResponse,  # type: ignore
+                            type_=typing.Optional[typing.Any],  # type: ignore
                             object_=_response.json(),
                         ),
                     ),
@@ -692,9 +761,9 @@ class AsyncRawEmbeddingsClient:
                 raise InternalServerError(
                     headers=dict(_response.headers),
                     body=typing.cast(
-                        ActualErrorResponse,
+                        typing.Optional[typing.Any],
                         parse_obj_as(
-                            type_=ActualErrorResponse,  # type: ignore
+                            type_=typing.Optional[typing.Any],  # type: ignore
                             object_=_response.json(),
                         ),
                     ),
@@ -703,9 +772,9 @@ class AsyncRawEmbeddingsClient:
                 raise ServiceUnavailableError(
                     headers=dict(_response.headers),
                     body=typing.cast(
-                        ActualErrorResponse,
+                        typing.Optional[typing.Any],
                         parse_obj_as(
-                            type_=ActualErrorResponse,  # type: ignore
+                            type_=typing.Optional[typing.Any],  # type: ignore
                             object_=_response.json(),
                         ),
                     ),
@@ -719,18 +788,19 @@ class AsyncRawEmbeddingsClient:
         self,
         *,
         tenant_id: str,
-        embeddings: typing.Optional[typing.Sequence[float]] = OMIT,
-        sub_tenant_id: typing.Optional[str] = OMIT,
-        max_chunks: typing.Optional[int] = OMIT,
+        sub_tenant_id: str,
+        query_embedding: typing.Sequence[float],
+        limit: typing.Optional[int] = OMIT,
+        filter_expr: typing.Optional[str] = OMIT,
+        output_fields: typing.Optional[typing.Sequence[str]] = OMIT,
         request_options: typing.Optional[RequestOptions] = None,
-    ) -> AsyncHttpResponse[EmbeddingsSearchData]:
+    ) -> AsyncHttpResponse[typing.List[RawEmbeddingSearchResult]]:
         """
         Find similar chunks using an embedding vector.
 
         Use this to retrieve the most similar chunk IDs to a single query embedding.
 
-
-        Expected outcome
+        Expected outcome:
         - You receive the closest chunk IDs with optional similarity scores.
 
         Parameters
@@ -738,30 +808,39 @@ class AsyncRawEmbeddingsClient:
         tenant_id : str
             Unique identifier for the tenant/organization
 
-        embeddings : typing.Optional[typing.Sequence[float]]
-            The embedding vector for search
-
-        sub_tenant_id : typing.Optional[str]
+        sub_tenant_id : str
             Optional sub-tenant identifier used to organize data within a tenant. If omitted, the default sub-tenant created during tenant setup will be used.
 
-        max_chunks : typing.Optional[int]
+        query_embedding : typing.Sequence[float]
+            Query embedding vector to search for
+
+        limit : typing.Optional[int]
+            Maximum number of results to return
+
+        filter_expr : typing.Optional[str]
+            Optional Milvus filter expression for additional filtering
+
+        output_fields : typing.Optional[typing.Sequence[str]]
+            Optional list of fields to return in results (default: chunk_id, source_id, metadata)
 
         request_options : typing.Optional[RequestOptions]
             Request-specific configuration.
 
         Returns
         -------
-        AsyncHttpResponse[EmbeddingsSearchData]
+        AsyncHttpResponse[typing.List[RawEmbeddingSearchResult]]
             Successful Response
         """
         _response = await self._client_wrapper.httpx_client.request(
-            "embeddings/search",
+            "embeddings/search-raw-embeddings",
             method="POST",
             json={
                 "tenant_id": tenant_id,
-                "embeddings": embeddings,
                 "sub_tenant_id": sub_tenant_id,
-                "max_chunks": max_chunks,
+                "query_embedding": query_embedding,
+                "limit": limit,
+                "filter_expr": filter_expr,
+                "output_fields": output_fields,
             },
             headers={
                 "content-type": "application/json",
@@ -772,9 +851,9 @@ class AsyncRawEmbeddingsClient:
         try:
             if 200 <= _response.status_code < 300:
                 _data = typing.cast(
-                    EmbeddingsSearchData,
+                    typing.List[RawEmbeddingSearchResult],
                     parse_obj_as(
-                        type_=EmbeddingsSearchData,  # type: ignore
+                        type_=typing.List[RawEmbeddingSearchResult],  # type: ignore
                         object_=_response.json(),
                     ),
                 )
@@ -783,9 +862,9 @@ class AsyncRawEmbeddingsClient:
                 raise BadRequestError(
                     headers=dict(_response.headers),
                     body=typing.cast(
-                        ActualErrorResponse,
+                        typing.Optional[typing.Any],
                         parse_obj_as(
-                            type_=ActualErrorResponse,  # type: ignore
+                            type_=typing.Optional[typing.Any],  # type: ignore
                             object_=_response.json(),
                         ),
                     ),
@@ -794,9 +873,9 @@ class AsyncRawEmbeddingsClient:
                 raise UnauthorizedError(
                     headers=dict(_response.headers),
                     body=typing.cast(
-                        ActualErrorResponse,
+                        typing.Optional[typing.Any],
                         parse_obj_as(
-                            type_=ActualErrorResponse,  # type: ignore
+                            type_=typing.Optional[typing.Any],  # type: ignore
                             object_=_response.json(),
                         ),
                     ),
@@ -805,9 +884,9 @@ class AsyncRawEmbeddingsClient:
                 raise ForbiddenError(
                     headers=dict(_response.headers),
                     body=typing.cast(
-                        ActualErrorResponse,
+                        typing.Optional[typing.Any],
                         parse_obj_as(
-                            type_=ActualErrorResponse,  # type: ignore
+                            type_=typing.Optional[typing.Any],  # type: ignore
                             object_=_response.json(),
                         ),
                     ),
@@ -816,9 +895,9 @@ class AsyncRawEmbeddingsClient:
                 raise NotFoundError(
                     headers=dict(_response.headers),
                     body=typing.cast(
-                        ActualErrorResponse,
+                        typing.Optional[typing.Any],
                         parse_obj_as(
-                            type_=ActualErrorResponse,  # type: ignore
+                            type_=typing.Optional[typing.Any],  # type: ignore
                             object_=_response.json(),
                         ),
                     ),
@@ -838,9 +917,9 @@ class AsyncRawEmbeddingsClient:
                 raise InternalServerError(
                     headers=dict(_response.headers),
                     body=typing.cast(
-                        ActualErrorResponse,
+                        typing.Optional[typing.Any],
                         parse_obj_as(
-                            type_=ActualErrorResponse,  # type: ignore
+                            type_=typing.Optional[typing.Any],  # type: ignore
                             object_=_response.json(),
                         ),
                     ),
@@ -849,9 +928,9 @@ class AsyncRawEmbeddingsClient:
                 raise ServiceUnavailableError(
                     headers=dict(_response.headers),
                     body=typing.cast(
-                        ActualErrorResponse,
+                        typing.Optional[typing.Any],
                         parse_obj_as(
-                            type_=ActualErrorResponse,  # type: ignore
+                            type_=typing.Optional[typing.Any],  # type: ignore
                             object_=_response.json(),
                         ),
                     ),
@@ -861,45 +940,59 @@ class AsyncRawEmbeddingsClient:
             raise ApiError(status_code=_response.status_code, headers=dict(_response.headers), body=_response.text)
         raise ApiError(status_code=_response.status_code, headers=dict(_response.headers), body=_response_json)
 
-    async def get_by_chunk_ids(
+    async def filter(
         self,
         *,
-        chunk_ids: typing.Sequence[str],
         tenant_id: str,
-        sub_tenant_id: typing.Optional[str] = OMIT,
+        sub_tenant_id: str,
+        source_id: typing.Optional[str] = OMIT,
+        chunk_ids: typing.Optional[typing.Sequence[str]] = OMIT,
+        output_fields: typing.Optional[typing.Sequence[str]] = OMIT,
+        limit: typing.Optional[int] = OMIT,
         request_options: typing.Optional[RequestOptions] = None,
-    ) -> AsyncHttpResponse[EmbeddingsGetData]:
+    ) -> AsyncHttpResponse[typing.List[RawEmbeddingSearchResult]]:
         """
         Retrieve embeddings for specific chunk IDs.
-
         Use this when you already know the chunk IDs and need their corresponding embeddings.
 
         Parameters
         ----------
-        chunk_ids : typing.Sequence[str]
-            The chunk IDs of the source you want to get embeddings for
-
         tenant_id : str
             Unique identifier for the tenant/organization
 
-        sub_tenant_id : typing.Optional[str]
+        sub_tenant_id : str
             Optional sub-tenant identifier used to organize data within a tenant. If omitted, the default sub-tenant created during tenant setup will be used.
 
+        source_id : typing.Optional[str]
+            Optional source ID to filter by (mutually exclusive with chunk_ids)
+
+        chunk_ids : typing.Optional[typing.Sequence[str]]
+            Optional list of chunk IDs to filter by (mutually exclusive with source_id)
+
+        output_fields : typing.Optional[typing.Sequence[str]]
+            Optional list of fields to return in results (default: chunk_id, source_id, metadata)
+
+        limit : typing.Optional[int]
+            Maximum number of results to return
+
         request_options : typing.Optional[RequestOptions]
             Request-specific configuration.
 
         Returns
         -------
-        AsyncHttpResponse[EmbeddingsGetData]
+        AsyncHttpResponse[typing.List[RawEmbeddingSearchResult]]
             Successful Response
         """
         _response = await self._client_wrapper.httpx_client.request(
-            "embeddings/retrieve_by_ids",
+            "embeddings/filter-raw-embeddings",
             method="POST",
             json={
-                "chunk_ids": chunk_ids,
                 "tenant_id": tenant_id,
                 "sub_tenant_id": sub_tenant_id,
+                "source_id": source_id,
+                "chunk_ids": chunk_ids,
+                "output_fields": output_fields,
+                "limit": limit,
             },
             headers={
                 "content-type": "application/json",
@@ -910,9 +1003,9 @@ class AsyncRawEmbeddingsClient:
         try:
             if 200 <= _response.status_code < 300:
                 _data = typing.cast(
-                    EmbeddingsGetData,
+                    typing.List[RawEmbeddingSearchResult],
                     parse_obj_as(
-                        type_=EmbeddingsGetData,  # type: ignore
+                        type_=typing.List[RawEmbeddingSearchResult],  # type: ignore
                         object_=_response.json(),
                     ),
                 )
@@ -921,9 +1014,9 @@ class AsyncRawEmbeddingsClient:
                 raise BadRequestError(
                     headers=dict(_response.headers),
                     body=typing.cast(
-                        ActualErrorResponse,
+                        typing.Optional[typing.Any],
                         parse_obj_as(
-                            type_=ActualErrorResponse,  # type: ignore
+                            type_=typing.Optional[typing.Any],  # type: ignore
                             object_=_response.json(),
                         ),
                     ),
@@ -932,9 +1025,9 @@ class AsyncRawEmbeddingsClient:
                 raise UnauthorizedError(
                     headers=dict(_response.headers),
                     body=typing.cast(
-                        ActualErrorResponse,
+                        typing.Optional[typing.Any],
                         parse_obj_as(
-                            type_=ActualErrorResponse,  # type: ignore
+                            type_=typing.Optional[typing.Any],  # type: ignore
                             object_=_response.json(),
                         ),
                     ),
@@ -943,9 +1036,9 @@ class AsyncRawEmbeddingsClient:
                 raise ForbiddenError(
                     headers=dict(_response.headers),
                     body=typing.cast(
-                        ActualErrorResponse,
+                        typing.Optional[typing.Any],
                         parse_obj_as(
-                            type_=ActualErrorResponse,  # type: ignore
+                            type_=typing.Optional[typing.Any],  # type: ignore
                             object_=_response.json(),
                         ),
                     ),
@@ -954,9 +1047,9 @@ class AsyncRawEmbeddingsClient:
                 raise NotFoundError(
                     headers=dict(_response.headers),
                     body=typing.cast(
-                        ActualErrorResponse,
+                        typing.Optional[typing.Any],
                         parse_obj_as(
-                            type_=ActualErrorResponse,  # type: ignore
+                            type_=typing.Optional[typing.Any],  # type: ignore
                             object_=_response.json(),
                         ),
                     ),
@@ -976,9 +1069,9 @@ class AsyncRawEmbeddingsClient:
                 raise InternalServerError(
                     headers=dict(_response.headers),
                     body=typing.cast(
-                        ActualErrorResponse,
+                        typing.Optional[typing.Any],
                         parse_obj_as(
-                            type_=ActualErrorResponse,  # type: ignore
+                            type_=typing.Optional[typing.Any],  # type: ignore
                             object_=_response.json(),
                         ),
                     ),
@@ -987,9 +1080,9 @@ class AsyncRawEmbeddingsClient:
                 raise ServiceUnavailableError(
                     headers=dict(_response.headers),
                     body=typing.cast(
-                        ActualErrorResponse,
+                        typing.Optional[typing.Any],
                         parse_obj_as(
-                            type_=ActualErrorResponse,  # type: ignore
+                            type_=typing.Optional[typing.Any],  # type: ignore
                             object_=_response.json(),
                         ),
                     ),
@@ -999,41 +1092,60 @@ class AsyncRawEmbeddingsClient:
             raise ApiError(status_code=_response.status_code, headers=dict(_response.headers), body=_response.text)
         raise ApiError(status_code=_response.status_code, headers=dict(_response.headers), body=_response_json)
 
-    async def create_collection(
-        self, *, tenant_id: str, request_options: typing.Optional[RequestOptions] = None
-    ) -> AsyncHttpResponse[EmbeddingsCreateCollectionData]:
+    async def delete(
+        self,
+        *,
+        tenant_id: str,
+        sub_tenant_id: typing.Optional[str] = None,
+        source_id: typing.Optional[str] = None,
+        chunk_ids: typing.Optional[typing.Union[str, typing.Sequence[str]]] = None,
+        request_options: typing.Optional[RequestOptions] = None,
+    ) -> AsyncHttpResponse[DeleteResult]:
         """
-        Create an embeddings collection for a tenant.
+        Delete embedding chunks by chunk ID.
 
-        Use this to initialize storage so you can index and query embeddings for the tenant.
+        Use this to remove specific chunks from your embeddings index when they are no longer
+        valid or should not appear in results.
 
         Parameters
         ----------
         tenant_id : str
             Unique identifier for the tenant/organization
 
+        sub_tenant_id : typing.Optional[str]
+            Optional sub-tenant ID for scoping deletion
+
+        source_id : typing.Optional[str]
+            Optional source ID to delete by (mutually exclusive with chunk_ids)
+
+        chunk_ids : typing.Optional[typing.Union[str, typing.Sequence[str]]]
+            Optional list of chunk IDs to delete (mutually exclusive with source_id)
+
         request_options : typing.Optional[RequestOptions]
             Request-specific configuration.
 
         Returns
         -------
-        AsyncHttpResponse[EmbeddingsCreateCollectionData]
+        AsyncHttpResponse[DeleteResult]
             Successful Response
         """
         _response = await self._client_wrapper.httpx_client.request(
-            "embeddings/create_tenant",
-            method="POST",
+            "embeddings/delete-raw-embeddings",
+            method="DELETE",
             params={
                 "tenant_id": tenant_id,
+                "sub_tenant_id": sub_tenant_id,
+                "source_id": source_id,
+                "chunk_ids": chunk_ids,
             },
             request_options=request_options,
         )
         try:
             if 200 <= _response.status_code < 300:
                 _data = typing.cast(
-                    EmbeddingsCreateCollectionData,
+                    DeleteResult,
                     parse_obj_as(
-                        type_=EmbeddingsCreateCollectionData,  # type: ignore
+                        type_=DeleteResult,  # type: ignore
                         object_=_response.json(),
                     ),
                 )
@@ -1042,9 +1154,9 @@ class AsyncRawEmbeddingsClient:
                 raise BadRequestError(
                     headers=dict(_response.headers),
                     body=typing.cast(
-                        ActualErrorResponse,
+                        typing.Optional[typing.Any],
                         parse_obj_as(
-                            type_=ActualErrorResponse,  # type: ignore
+                            type_=typing.Optional[typing.Any],  # type: ignore
                             object_=_response.json(),
                         ),
                     ),
@@ -1053,9 +1165,9 @@ class AsyncRawEmbeddingsClient:
                 raise UnauthorizedError(
                     headers=dict(_response.headers),
                     body=typing.cast(
-                        ActualErrorResponse,
+                        typing.Optional[typing.Any],
                         parse_obj_as(
-                            type_=ActualErrorResponse,  # type: ignore
+                            type_=typing.Optional[typing.Any],  # type: ignore
                             object_=_response.json(),
                         ),
                     ),
@@ -1064,9 +1176,9 @@ class AsyncRawEmbeddingsClient:
                 raise ForbiddenError(
                     headers=dict(_response.headers),
                     body=typing.cast(
-                        ActualErrorResponse,
+                        typing.Optional[typing.Any],
                         parse_obj_as(
-                            type_=ActualErrorResponse,  # type: ignore
+                            type_=typing.Optional[typing.Any],  # type: ignore
                             object_=_response.json(),
                         ),
                     ),
@@ -1075,9 +1187,9 @@ class AsyncRawEmbeddingsClient:
                 raise NotFoundError(
                     headers=dict(_response.headers),
                     body=typing.cast(
-                        ActualErrorResponse,
+                        typing.Optional[typing.Any],
                         parse_obj_as(
-                            type_=ActualErrorResponse,  # type: ignore
+                            type_=typing.Optional[typing.Any],  # type: ignore
                             object_=_response.json(),
                         ),
                     ),
@@ -1097,9 +1209,9 @@ class AsyncRawEmbeddingsClient:
                 raise InternalServerError(
                     headers=dict(_response.headers),
                     body=typing.cast(
-                        ActualErrorResponse,
+                        typing.Optional[typing.Any],
                         parse_obj_as(
-                            type_=ActualErrorResponse,  # type: ignore
+                            type_=typing.Optional[typing.Any],  # type: ignore
                             object_=_response.json(),
                         ),
                     ),
@@ -1108,9 +1220,9 @@ class AsyncRawEmbeddingsClient:
                 raise ServiceUnavailableError(
                     headers=dict(_response.headers),
                     body=typing.cast(
-                        ActualErrorResponse,
+                        typing.Optional[typing.Any],
                         parse_obj_as(
-                            type_=ActualErrorResponse,  # type: ignore
+                            type_=typing.Optional[typing.Any],  # type: ignore
                             object_=_response.json(),
                         ),
                     ),