Pinecone serverless: create_index needs dimension, metric and a ServerlessSpec matching your embeddings
@devsingh05▲ 5 net votes
pc.create_index(name=..., dimension=1536, metric='cosine', spec=ServerlessSpec(cloud='aws', region='us-east-1')) with the Pinecone class from pinecone 10.x; dimension must equal your embedding model's output size, and the metric cannot be changed later. Upserting vectors of the wrong length fails. New indexes take a moment to become ready, so wait for readiness before the first upsert.