generate

Same renderer as /api/qr/generate, but you describe what you want in a sentence and the engine picks the shapes, colors and finder eyes.

There are three variants of the AI endpoint — pick the one that matches your auth.

Marketplace — bring your own LLM key

POST https://api.qr-branding.com/api/qr/ai/generate

curl -X POST https://api.qr-branding.com/api/qr/ai/generate \
  -H "Content-Type: application/json" \
  -H "X-RapidAPI-Proxy-Secret: <secret>" \
  -d '{
    "content": "https://yourbrand.com",
    "prompt": "art deco gold leaf, geometric chevrons, ivory background",
    "llmProvider": "openai",
    "llmApiKey": "sk-...",
    "creativity": 0.6,
    "strictScannability": true,
    "stylePreset": "luxury"
  }'

You supply the LLM credentials. The engine validates the resulting config, runs the scan guarantee, and renders the QR.

Marketplace — managed AI, no LLM key

curl -X POST https://api.qr-branding.com/api/qr/ai/generate-managed \
  -H "Content-Type: application/json" \
  -H "X-RapidAPI-Proxy-Secret: <secret>" \
  -d '{
    "content": "https://yourbrand.com",
    "prompt": "art deco gold leaf, geometric chevrons, ivory background",
    "creativity": 0.6,
    "strictScannability": true,
    "stylePreset": "luxury"
  }'

No llmProvider / llmApiKey needed. The server uses its own models with a Groq → Gemini → OpenAI fallback chain. The response includes "managed": true, and the call counts towards your plan's Managed AI quota.

Consumer — server-side LLM key

POST https://qr-branding.com/api/qr/ai/generate/server

curl -X POST https://qr-branding.com/api/qr/ai/generate/server \
  -H "Content-Type: application/json" \
  -d '{
    "content": "https://yourbrand.com",
    "prompt": "art deco gold leaf, geometric chevrons, ivory background",
    "stylePreset": "luxury",
    "creativity": 0.6
  }'

No llmProvider / llmApiKey needed. Server tries Groq (gpt-oss-120b, then gpt-oss-20b) → OpenAI → Gemini in order. One AI credit is deducted from your balance on success; on failure no credit is consumed.

Code samples

Request body

Field	Type	Default	Notes
`content`	string	required	What to encode (URL / text).
`prompt`	string	required	Plain language description, ≤ 10000 chars.
`stylePreset`	string	`luxury`	One of `minimal artistic corporate playful luxury tech nature`.
`creativity`	number	`0.5`	`0–1`. Affects LLM temperature + boldness of design choices.
`strictScannability`	boolean	`true`	When on, the engine auto-fixes any scannability issue. When off, it warns.
`logoUrl`	string	—	Embedded center, modules cleared beneath, ECC raised to H.
`logoBase64`	string	—	Same as above but inline base64.
`backgroundImageUrl`	string	—	Soft background image (blended behind the QR).
`centerText`	object	—	`{ text, color?, backgroundColor?, fontSize? }`. Banner across the QR.
`ctaText`	string	—	Caption below.
`fileFormat`	string	`png`	`png jpg svg pdf webp`.
`outputWidth`	number	`800`	Pixels.
`responseType`	string	`base64`	`base64` or `binary`.

Marketplace endpoint additionally requires:

Field	Type	Notes
`llmProvider`	string	`openai anthropic google mistral cohere groq xai deepseek qwen local`
`llmApiKey`	string	Your provider key
`llmModel`	string	Optional — defaults to the provider's default model (see `/api/qr/ai/providers`); any model ID the provider offers is accepted

Response

{
  "success": true,
  "qrBase64": "iVBORw0KGgo…",
  "format": "png",
  "contentType": "image/png",
  "sizeBytes": 38291,
  "generatedConfig": { "moduleShape": "diamond", "finderOuterShape": "octagon", "...": "..." },
  "aiExplanation": "Diamond modules and octagonal finders for the geometric art deco feel; gold-on-ivory keeps contrast at 6.4:1.",
  "llmProviderUsed": "openai",
  "llmModelUsed": "gpt-6-luna"
}

generatedConfig is the full QrConfig the AI produced — useful if you want to take that config into Manual mode and tweak it by hand.