Embeddings

Text Embeddings

Looking for the OpenAI wire path? POST /v1/embeddings ($0.002) speaks the standard OpenAI embeddings format — batch up to 64 inputs / 16k chars per request, dimensions and encoding_format pass through, and byte-identical repeats within 10 minutes are served free automatically (embeddings are deterministic; opt out with cache: false). See LLM Gateway (OpenAI /v1). The /api/embed* tools below remain available as the simpler custom-JSON shape.

Two tiers of text embedding generation, paywalled via x402. Send text, get back a vector for semantic search, RAG, clustering, or similarity. The operator's OPENAI_API_KEY handles upstream auth.

Tiers

Endpoint	Price	Model	Dimensions	Text cap
POST /api/embed	$0.005	text-embedding-3-small	1,536	32,000 chars
POST /api/embed-large	$0.01	text-embedding-3-large	3,072	32,000 chars

Both tiers are wallet-only — every call burns real upstream embedding credit. See Security Model.

Request / Response

// Request
{ "text": "Agent402 is an open-source x402 tool server." }

// Response
{ "model": "text-embedding-3-small", "provider": "openai",
  "embedding": [0.0023, -0.0091, 0.0152, ...],
  "dimensions": 1536,
  "usage": { "total_tokens": 12 } }

When to use which tier

• embed ($0.005) — good enough for most RAG, search, and clustering. 1,536 dimensions.
• embed-large ($0.01) — higher accuracy for precision-critical applications. 3,072 dimensions.

See also

• LLM Proxy Gateway — text inference
• Paying with x402 — the USDC payment flow