From key to first request in minutes.

Quickstart

Make your first request

Three steps: create a key, set the base URL, send a request. The only thing that changes between models is the value of the model field.

1. 1
   Create an API key

   Generate one in API Keys and store it as an environment variable.

2. 2
   Point your client at PrByCode

   Set the base URL to https://api.prbycode.com/v1.

3. 3
   Send a request

   Pick any model ID from the catalog and send a standard completion request.

Authentication

Send your key as a bearer token on every request. Keys are scoped to a single workspace, and each key can be revoked independently of the others.

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

Choosing a model

Models are selected per request. A request that omits model uses the default configured in the console; if that model fails, the fallback chain is tried in order.

Fastest

deepseek-v4.1-flash

Lowest cost

deepseek-v4.1-flash

Longest context

claude-sonnet-5

Vision

gpt-5.6-luna

The model catalog lists context window, capabilities and per-token pricing for every route.

Chat completions

The primary endpoint. It accepts the standard OpenAI request body, including tools, response_format and stream.

curl https://api.prbycode.com/v1/chat/completions \
  -H "Authorization: Bearer $PRBYCODE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "glm-5.3-flash",
    "messages": [{ "role": "user", "content": "Hello" }]
  }'

Anthropic messages

Clients built against the Anthropic messages shape can point at PrByCode without changes. System prompts are passed as a top-level field rather than a message.

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" }]
  }'

Streaming

Set stream to true to receive server-sent events. Each chunk carries an incremental delta, and the stream ends with a [DONE] sentinel.

data: {"choices":[{"delta":{"role":"assistant","content":"Hello"}}]}

data: {"choices":[{"delta":{"content":" from"}}]}

data: {"choices":[{"delta":{"content":" PrByCode."}}]}

data: {"choices":[{"delta":{},"finish_reason":"stop"}]}

data: [DONE]

Usage counters arrive in the final chunk, so a streamed request still reports token consumption in Activity.

List models

Return the models currently routable with your key, in the standard OpenAI list shape.

curl https://api.prbycode.com/v1/models \
  -H "Authorization: Bearer $PRBYCODE_API_KEY"

Error handling

Failures use conventional HTTP status codes with a JSON body. Every response carries a request ID you can search for in the console.

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

Rate limits and quota

Requests are limited per key, and token consumption is metered against your plan allowance. Both are visible in the console before you hit them.

Rate limit headers

x-ratelimit-remaining-requests

Retry after

retry-after

Exhausted allowance

HTTP 429 until the period renews

Live view

Usage

Client compatibility

Any client that accepts a custom OpenAI-compatible base URL works without further changes.

• OpenAI official SDKs for Node.js, Python, Go, Java and .NET
• Framework integrations that accept a configurable base URL
• Editors and CLI tools with a custom endpoint setting
• Anthropic-compatible clients against /v1/messages