Vespa Cloud roadmap

This page lists large upcoming features in Vespa Cloud. See also the Vespa roadmap.

Launching soon

• Pre-trained machine-learned models available for embedding documents and queries in dense vector spaces for dense-retrieval using Vespa’s nearest neighbor search query operator support. These models also enable hybrid retrieval combining sparse and dense representations.

In progress

• Choose to deploy in GCP zones in addition to AWS.
• Allow applications to deploy applications running on nodes in their own AWS accounts.
• Support Graviton nodes in AWS.
• Pre-trained machine-learned models available for ranking.

Priority backlog

• Support GPUs in container clusters.
• Auto-sizing: Specify required capacity and performance rather than node resources.