Pinecone launches Nexus (agent knowledge engine) in public preview
Pinecone Nexus, a knowledge engine for AI agents that offloads retrieval, assembly, and reasoning from the LLM to a dedicated layer, entered public preview. Access is request-only with no public price yet.
Pinecone sold Database (serverless), Inference, and Assistant.
Adds Nexus (public preview); egress now documented as a 4th serverless billing meter ($0.10/GB, 100 GB/mo included).
Pinecone announced Nexus, described as a knowledge engine for agents that “offloads retrieval, assembly, and reasoning from the LLM to a dedicated knowledge layer — done once when your data changes, not on every agent call.” It is in public preview and is request-access only, with no public pricing at launch.
Alongside the launch, Pinecone’s serverless pricing now documents a fourth usage meter — egress — billed at $0.10/GB with a monthly allowance (1 GB Free, 10 GB Builder, 100 GB Standard/Enterprise), sitting beside the existing read-unit, write-unit, and storage meters. Core Database plan rates (RU/WU/storage) and the Starter/Builder/Standard/Enterprise + BYOC structure are unchanged, and the one-time $250 (1 TB) bulk-import credit for Standard/Enterprise was extended to 2026-08-31.
Serverless pricing now documents four usage meters — read units, write units, storage, and egress ($0.10/GB, 100 GB/mo included on Standard/Enterprise). Pinecone Nexus (agent knowledge engine) launched in public preview. Core RU/WU/storage plan rates unchanged; the $250 (1 TB) bulk-import credit was extended to 2026-08-31.