API Overview

AnyPath exposes a small public gateway surface over platform-priced shared AI capacity.

Base URLs

Protocol	Base URL	Use when
OpenAI-compatible	https://api.anypath.dev/v1	Chat Completions, Responses, model discovery, and images
Anthropic Messages	https://api.anypath.dev	Native Anthropic Messages request shapes
Gemini /v1beta	https://api.anypath.dev/v1beta	Gemini-compatible REST request shapes

Authentication

Protocol	Supported auth
OpenAI-compatible	Authorization: Bearer sk-...
Anthropic Messages	Authorization: Bearer sk-... or x-api-key: sk-...
Gemini /v1beta	Authorization: Bearer sk-..., x-goog-api-key: sk-..., or ?key=sk-...

See Authentication for key handling and error responses.

Available endpoints

OpenAI-compatible

Endpoint	Method	Description
/v1/models	GET	Model discovery for the current API key and route policy
/v1/routes	GET	Public route catalog for the gateway contract
/v1/chat/completions	POST	Chat Completions request shape
/v1/responses	POST	Responses request shape
/v1/images/generations	POST	Image generation
/v1/images/edits	POST	Image editing

Anthropic Messages

Endpoint	Method	Description
/v1/messages	POST	Claude-native Messages request shape
/v1/messages/count_tokens	POST	Anthropic-compatible token count request shape

Gemini /v1beta

Endpoint	Method	Description
/v1beta/models	GET	List Gemini-compatible models
/v1beta/models/{model}	GET	Get Gemini-compatible model metadata
/v1beta/models/{model}:generateContent	POST	Generate content
/v1beta/models/{model}:streamGenerateContent	POST	Stream generated content
/v1beta/models/{model}:countTokens	POST	Gateway token estimate for Gemini-compatible request bodies

Rate limits

The account tier sets the account policy, while each API key can narrow model access, route policy, budget, RPM, and TPM. The API key route policy controls whether AnyPath uses intelligent routing or stays inside a specified provider. Read /v1/models and Console limits for the current effective RPM/TPM values.

When a request exceeds its limit, the gateway returns 429 with a Retry-After hint.

See Rate Limits for retry behavior.

Error response format

OpenAI-compatible gateway errors use this envelope:

{
  "error": {
    "message": "Invalid API key",
    "type": "auth_invalid",
    "code": "auth_invalid"
  }
}

At runtime type equals code. See Error Codes for the full registry.

Routing and settlement

Requests use routes available to the API key:

• Shared capacity settles in USDT after successful requests.
• Shared-route pricing is derived from official model price and the current provider-level multiplier.
• Route policy and availability determine which models the key can use.