Simplifying Vector Embeddings with Pinecone Integrated Inference Capabilities
Oct 9, 2025John Ward
Summary
This post details a workaround for the 40KB metadata limit when using Pinecone's `upsert_records` method with integrated inference. The issue arises because `upsert_records` automatically includes the full text field in metadata. The solution involves directly calling the Pinecone Inference API to generate embeddings, then upserting these embeddings with controlled metadata, thus avoiding the automatic inclusion of the full text and circumventing the metadata size constraint.
More from the Integrated Inference for Embeddings trail
- Retrieval Inference for scale and performance
- Introducing Pinecone Rerank V0
- Introducing integrated inference: Embed, rerank, and retrieve your data with a single API
- Introducing reranking to Pinecone Inference to simplify building accurate AI
- Introducing Pinecone Inference to streamline your AI workflow