Vector DB & Managed RAG

What it is

Vector DB stores embeddings (768–4096 dimensions) and serves k-NN queries from every PoP. Managed RAG is the turnkey pipeline on top: drop in documents, choose a chunking strategy, pick an embedding model from the LLM Gateway, and you get a query endpoint that returns top-k passages.

When to use it

RAG over a knowledge base without operating your own pipeline.
Semantic search backing a product feature.
Embedding storage that lives next to your edge inference.

Quickstart

asc edge vector create kb --dims 1536
asc edge rag ingest kb ./docs/*.md --chunk-size 800 --overlap 100 --embed text-embedding-3-small
asc edge rag query kb 'what is the rate limit?' --top-k 5

client.edge.vector.create('kb', dims=1536)
client.edge.rag.ingest('kb', files=['./docs/a.md'], chunk_size=800, overlap=100, embed='text-embedding-3-small')
hits = client.edge.rag.query('kb', 'what is the rate limit?', top_k=5)

const hits = await client.edge.rag.query('kb', { question: 'what is the rate limit?', topK: 5 });

Limits & quotas

Limit	Default	Burst	Notes
Dimensions stored	$0.05 / M dim-mo	—
Vector queries	$0.04 / M	—
RAG embedding tokens	$0.04 / M	—	BYOK supported
RAG retrieval queries	$1.00 / 100K	—
Max dimensions	4,096	—

Pricing

See pricing. Pay-as-you-go, billed monthly via Stripe.

API surface

POST /v1/edge/vector
POST /v1/edge/rag/{kb}/ingest
POST /v1/edge/rag/{kb}/query

Required scope(s): edge:write. See Scopes.

Security

All access is authenticated and scoped. See Auth & scopes and Network controls.