One endpoint, a familiar surface.

Base URL

All requests are made against a single host. The version segment is part of the base URL.

Authentication

Send the key as a bearer token. The Anthropic-shaped endpoint also accepts the key in an x-api-key header.

Authorization: Bearer prby_live_••••••••••••

Conventions

• Request and response bodies are JSON encoded as UTF-8.
• Timestamps in responses are ISO 8601 UTC.
• Token counts follow the routed model's own tokenizer.
• Every response includes an x-prbycode-request-id header.
• Unknown fields in a request body are ignored rather than rejected.

POST /v1/chat/completions

Create a model response from a list of messages. This is the endpoint most clients use.

curl https://api.prbycode.com/v1/chat/completions \
  -H "Authorization: Bearer $PRBYCODE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "claude-sonnet-5",
    "messages": [
      { "role": "system", "content": "You are a concise assistant." },
      { "role": "user", "content": "Summarise the request lifecycle." }
    ],
    "max_tokens": 512,
    "temperature": 0.7,
    "stream": false
  }'

POST /v1/messages

The Anthropic-compatible shape. Differences from the chat completions endpoint are the API key header, the required max_tokens field, and system as a top-level field.

curl https://api.prbycode.com/v1/messages \
  -H "x-api-key: $PRBYCODE_API_KEY" \
  -H "anthropic-version: 2023-06-01" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "claude-sonnet-5",
    "max_tokens": 1024,
    "system": "You are a concise assistant.",
    "messages": [{ "role": "user", "content": "Hello" }]
  }'

GET /v1/models

Return the models routable with the presented key.

{
  "object": "list",
  "data": [
    {
      "id": "claude-sonnet-5",
      "object": "model",
      "owned_by": "anthropic",
      "context_window": 1000000
    },
    {
      "id": "glm-5.3-flash",
      "object": "model",
      "owned_by": "zhipu ai",
      "context_window": 256000
    }
  ]
}

GET /v1/quota

Inspect plan status, token allowance and current rate limit standing without opening the console.

{
  "object": "quota",
  "plan": "Monthly",
  "period": {
    "start": "2026-09-18",
    "renews": "2026-10-18"
  },
  "tokens": {
    "used": 412884000,
    "total": 1000000000
  },
  "rate_limit": {
    "limit": 600,
    "remaining": 574,
    "reset_at": "2026-09-20T04:00:00Z"
  }
}

Request parameters

Fields accepted by /v1/chat/completions. Availability depends on the routed model — see the parameters tab on any model page.

• modelYesstring · Model ID from the catalog. Falls back to the default route when omitted.
• messagesYesarray · Conversation history as { role, content } objects.
• streamNoboolean · Return server-sent events instead of a single JSON body.
• temperatureNonumber · Sampling temperature. Model defaults apply when omitted.
• top_pNonumber · Nucleus sampling probability mass.
• max_tokensNointeger · Upper bound on completion tokens.
• toolsNoarray · Function definitions the model may call.
• tool_choiceNostring | object · Whether and which tool the model must call.
• response_formatNoobject · Request a JSON response shape.
• stopNostring | array · Sequences that terminate generation.
• userNostring · Stable end-user identifier for abuse attribution.

Response object

A non-streamed response is a standard completion object with usage counters attached.

{
  "id": "req_8f21c0d4a7b2",
  "object": "chat.completion",
  "created": 1789000000,
  "model": "claude-sonnet-5",
  "choices": [
    {
      "index": 0,
      "finish_reason": "stop",
      "message": {
        "role": "assistant",
        "content": "Requests enter the gateway, are validated and metered, then routed to a healthy model."
      }
    }
  ],
  "usage": {
    "prompt_tokens": 38,
    "completion_tokens": 24,
    "total_tokens": 62
  }
}

Errors

Failures use conventional status codes and a JSON body carrying a machine-readable type plus a request ID.

{
  "error": {
    "type": "rate_limit_error",
    "code": 429,
    "message": "Rate limit reached for this API key. Retry after the window resets.",
    "request_id": "req_8f21c0d4a7b2"
  }
}

Rate limits

Limits are enforced per key. Response headers report the remaining budget so a client can back off before it is refused.

Remaining requests

x-ratelimit-remaining-requests

Remaining tokens

x-ratelimit-remaining-tokens

Reset time

x-ratelimit-reset

Back-off hint

retry-after