What is the difference between Pinecone pods and serverless?

Pinecone pods are provisioned dedicated resources with fixed capacity, billed per hour regardless of usage, giving predictable performance and low cold-start latency. Serverless indexes auto-scale compute and storage based on actual load, billed per read unit, write unit, and storage, with no infrastructure to manage. Serverless separates storage from compute, using object storage for durability, which cuts cost dramatically for low-traffic or spiky workloads. Pods remain available for existing customers with high steady throughput, but serverless is the recommended default for new Pinecone projects.