microsoft/semantic-kernel · error · ServiceInvalidRequestError

Limit must be less than or equal to {MAX_QUERY_WITHOUT_METAD

Error message

Limit must be less than or equal to {MAX_QUERY_WITHOUT_METADATA_BATCH_SIZE}

What it means

Raised in PineconeMemoryStore.get_nearest_matches() when limit exceeds MAX_QUERY_WITHOUT_METADATA_BATCH_SIZE (10000). ServiceInvalidRequestError (a subclass of ServiceResponseException) fires after the collection-existence check but before the query. This enforces Pinecone's documented top_k ceiling for queries without metadata.

Source

Thrown at python/semantic_kernel/connectors/memory_stores/pinecone/pinecone_memory_store.py:354

        """Gets the nearest matches to an embedding using cosine similarity.

        Args:
            collection_name (str): The name of the collection to get the nearest matches from.
            embedding (ndarray): The embedding to find the nearest matches to.
            limit (int): The maximum number of matches to return.
            min_relevance_score (float): The minimum relevance score of the matches. (default: {0.0})
            with_embeddings (bool): Whether to include the embeddings in the results. (default: {False})

        Returns:
            List[Tuple[MemoryRecord, float]]: The records and their relevance scores.
        """
        if not await self.does_collection_exist(collection_name):
            raise ServiceResourceNotFoundError(f"Collection '{collection_name}' does not exist")

        collection = self.pinecone.Index(collection_name)

        if limit > MAX_QUERY_WITHOUT_METADATA_BATCH_SIZE:
            raise ServiceInvalidRequestError(
                "Limit must be less than or equal to " + f"{MAX_QUERY_WITHOUT_METADATA_BATCH_SIZE}"
            )
        if limit > MAX_QUERY_WITH_METADATA_BATCH_SIZE:
            query_response = collection.query(
                vector=embedding.tolist(),
                top_k=limit,
                include_values=False,
                include_metadata=False,
            )
            keys = [match.id for match in query_response.matches]
            fetch_response = await self.__get_batch(collection_name, keys, with_embeddings)
            vectors = fetch_response.vectors
            for match in query_response.matches:
                vectors[match.id].update(match)
            matches = [vectors[key] for key in vectors]
        else:
            query_response = collection.query(
                vector=embedding.tolist(),

View on GitHub (pinned to c028a0c7dc)

Solutions

1. Reduce limit to <= 10000 (and <= 1000 when metadata is requested).
2. Cap user-supplied limit at the boundary before calling.
3. Paginate via multiple smaller queries if more results are needed.
4. Migrate to PineconeStore + Collection.

Example fix

// before
matches = await store.get_nearest_matches("my_col", emb, limit=50000, with_embeddings=False)

// after
limit = min(user_limit, 10000)
matches = await store.get_nearest_matches("my_col", emb, limit=limit, with_embeddings=False)

from semantic_kernel.connectors.memory_stores.pinecone.pinecone_memory_store import MAX_QUERY_WITHOUT_METADATA_BATCH_SIZE
limit = min(limit, MAX_QUERY_WITHOUT_METADATA_BATCH_SIZE)
return await store.get_nearest_matches(collection_name, embedding, limit, with_embeddings)

def is_valid_query_limit(limit: int, with_metadata: bool) -> bool:
    return limit <= (1000 if with_metadata else 10000)

from semantic_kernel.exceptions import ServiceInvalidRequestError
try:
    return await store.get_nearest_matches(collection_name, embedding, limit, with_embeddings)
except ServiceInvalidRequestError as e:
    raise ValueError(f"limit too large for Pinecone: {e}") from e

When it happens

Trigger: Calling get_nearest_matches(..., limit=N) with N > 10000. Note the code checks the no-metadata ceiling first even when with_embeddings/with_metadata is requested; a higher ceiling (MAX_QUERY_WITH_METADATA_BATCH_SIZE=1000) is checked next.

Common situations: Caller passes an unbounded 'top N' from user input; pagination limit copied from a different backend; desire to retrieve the whole index in one call.

Related errors

• Dimensionality of {default_dimensionality} exceeds the maxim
• Failed to create Pinecone settings.
• Dimensionality of {dimension_num} exceeds the maximum allowe
• Collection '{collection_name}' does not exist
• Error upserting record: {upsert_response.message}

AI-assisted analysis of microsoft/semantic-kernel@c028a0c7dc (2026-08-13). Data as JSON: /api/errors/14aa44235c2a4028. Report an issue: GitHub.