Forge
Sign in
Build

API reference

Forge implements the OpenAI v1 API. Every endpoint below is a drop-in replacement — point any OpenAI SDK at our base URL.

Base URL

texttext
https://api.cytokenhub.com/v1

Chat completions

POST /v1/chat/completions

jsonjson
{
  "model": "anthropic/claude-4.5-sonnet",
  "messages": [{ "role": "user", "content": "Hello" }],
  "stream": false,
  "temperature": 0.7,
  "max_tokens": 1024,
  "tools": [],
  "response_format": { "type": "json_object" }
}

Forge-specific headers

  • X-Forge-Route — cheapest, fastest, or a comma-separated fallback chain.
  • X-Forge-Channel — pin a specific upstream channel code.
  • X-Forge-Budget — max cost in USD for this call.

Embeddings

POST /v1/embeddings

jsonjson
{
  "model": "openai/text-embedding-3-large",
  "input": ["one", "two", "three"]
}

Images

POST /v1/images/generations

jsonjson
{
  "model": "openai/gpt-image-1",
  "prompt": "a cat in a top hat",
  "size": "1024x1024",
  "n": 1
}

Audio

  • POST /v1/audio/speech — text-to-speech
  • POST /v1/audio/transcriptions — speech-to-text

Metadata

  • GET /v1/meta/model/list — every callable model + pricing
  • GET /v1/meta/endpoint/list — every gateway endpoint
  • GET /v1/meta/supplier/list — providers
  • GET /v1/meta/category/tree/all — category taxonomy

Account

  • GET /v1/apikey/balance/{code} — remaining quota
  • GET /v1/apikey/cost/{code} — spend
  • GET /v1/apikey/qps/{code} — live QPS
  • GET /v1/metrics — request timeseries