Embeddings

Routed across OpenAI, Mixedbread, and DeepInfra. Accepts one text or a batch of up to 100. Billing is per 1k input tokens.

Pin a provider for indexes

Vectors from different models are not interchangeable. For one-off similarity checks, routing freely is fine — but when building a persistent index, pin a single provider with routing.providers or a priority order in your routing preferences, and note the model field returned with every response.

Request

POST /v1/tools/embed
{
  "input": ["how do durable objects bill?", "cloudflare pricing model"],
  "routing": { "providers": ["openai-embed"] }
}

Result

{
  "result": {
    "embeddings": [[0.0121, -0.0334, …], [0.0087, 0.0412, …]],
    "dimensions": 1536,
    "model": "text-embedding-3-small",
    "tokens": 14
  }
}