Build with every frontier model through one stable API.

Quickstart

If your application already uses the OpenAI SDK, point the client at the KancyAI base URL and replace the API key. You can then choose any provider-specific model such as gpt-4o or claude-3-opus.

curl https://api.kancy.ai/v1/chat/completions \
  -H "Authorization: Bearer $KANCY_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gpt-4o",
    "messages": [
      {"role": "user", "content": "Draft a concise API launch note"}
    ]
  }'

Authentication

Every request uses bearer-token authentication. Create separate API keys for production, staging, and local development so usage and permissions remain easy to audit.

Authorization: Bearer $KANCY_API_KEY

Chat Completions

Use chat completions for reasoning, summarization, extraction, tool planning, and agent workflows. The endpoint supports provider-specific models — switch between them by changing the model field in each request.

POST /v1/chat/completions

{
  "model": "claude-3-opus",
  "messages": [
    {"role": "system", "content": "You are a precise technical assistant."},
    {"role": "user", "content": "Compare these two vendor contracts."}
  ],
  "temperature": 0.2
}

Image Generation

Generate product visuals, article covers, marketing assets, and design references through a unified media endpoint. Choose any image model such as Google, OpenAI Image, MiniMax Image, or ByteDance Image by specifying its model ID.

POST /v1/images/generations

{
  "model": "google-image",
  "prompt": "A clean enterprise API dashboard, realistic product UI",
  "size": "1536x1024"
}

Video Generation

Video generation is asynchronous. Create a job, store the returned job ID, then poll the job status or configure a webhook once your workspace enables callbacks.

POST /v1/videos/generations

{
  "model": "kling-video",
  "prompt": "A calm walkthrough of a global model gateway network",
  "duration": 6,
  "aspect_ratio": "16:9"
}

Model Switching

KancyAI does not perform automatic dynamic routing. Instead, you stay in full control: every request runs against the exact model you specify. Switch between providers and model families at any time by changing the model field — no code changes, no provider SDK swaps, no fallback configuration required.

Model IDs

Use direct model IDs to choose the exact provider and model for each request. Switch between models at any time by changing the model field — KancyAI keeps the request and response format identical across every provider.

Errors

KancyAI returns consistent HTTP status codes across providers. Provider-specific details are normalized into stable error categories whenever possible.

Rate Limits and Usage

Limits can be configured per workspace, project, API key, and modality. Enterprise workspaces can set spend caps, per-team quotas, and alert thresholds for production usage.