# Embeddings API

> OpenAI-compatible text embeddings: E5-small-v2 (384 dims) on Mojo CPU kernels, $0.02 per million tokens

MCP: https://embed.mojojojo.cc/mcp (JSON-RPC over POST). OpenAPI: https://embed.mojojojo.cc/openapi.json.
Auth: `Authorization: Bearer mj_live_...`; create a key at https://mojojojo.cc/keys. Usage is billed from mojojojo credits.

## Pricing

- Embedding tokens: $0.00002 per 1k tokens (1000 free per month)

## Endpoints

- `POST /v1/embeddings` (MCP tool `embed_text`): Embed one or more texts with E5-small-v2 (384-dim, L2-normalised). Use input_type=query for search queries and passage for documents.

## Limits and measured results

- input_type is "query" or "passage" (E5 prefixes are added for you); default passage.
- Vectors are L2-normalised, 384 dimensions; up to 512 inputs of 20,000 characters per request.
- Matches the PyTorch reference to 2e-7 max abs difference (cosine 0.9999999).
- Speed: 0.40x PyTorch at batch 1 up to 0.93x at batch 200 on 2 pinned cores. It is a convenience and price play, not a latency win.