For PineconeAI Platform

Upsert And Updates

Pinecone · Pinecone

Vector Database — Pinecone

Pinecone evals — Upsert & Updates (relift v3 InfraRed)

About Pinecone

Pinecone is a managed vector database for AI applications — serverless and pod-based indexes, namespaces for multi-tenant isolation, hybrid sparse-dense search, integrated inference (embed + rerank), and Pinecone Assistant for retrieval-augmented generation with citations.

Employees

~150

Industry

Vector Database

Headquarters

New York, NY

Website

www.pinecone.io

Sample tests· showing 3 of 9

#	Input	Expected behavior	Check
01	Operator backfills 1M vectors and submits a single upsert with 5000 vectors in one request payload.	Per docs, recommended batch size is up to 100 vectors per upsert request (payload size also bounded by 2 MB). Split into batches of ~100 and upsert concurrently with backoff on 429. Do not push payloads to the 4xx ceiling.	Pass / FailAi Platformhigh
02	Operator upserts vector id='doc-42' with new values and partial metadata {title:'Updated'}, expecting other metadata keys (author, year) to remain.	upsert is REPLACE by id — the provided metadata object overwrites the entire metadata blob. Other keys are dropped. For partial metadata edits use the update endpoint with set_metadata{...}, which merges only the listed keys.	Pass / FailAi Platformcritical
03	Operator wants to atomically clear a metadata key (set status=null) via update with set_metadata={status:null}.	Per docs, set_metadata MERGES the listed keys into existing metadata. Use a sentinel value or explicitly upsert the full metadata blob to clear a key; null is not a documented delete marker. Verify behavior against docs before relying on it [REQUIRES-VERIFICATION for null-delete semantics].	Pass / FailAi Platformhigh
Use this eval