Every AI model.
One API key. One wallet.
Generate images, video and audio from dozens of models through a single API — pay per generation in ₹ or $, and never pay for a failed run.
Text to image · OpenAI
GPT Image 2.5 Flare · Text to Image
Generate images from text with GPT Image 2.5 Flare. 1K, 2K or 4K, transparent backgrounds supported.
from $0.036/image
Models
Popular right now
Why ANKY.AI
Everything you need to ship with AI models
One key, one wallet
Every model behind the same API and the same prepaid balance — no per-provider accounts or cards.
Failed runs cost nothing
We hold the price when a task starts and only charge when it succeeds. Failures are refunded automatically.
Pay in ₹ or $
INR top-ups via UPI, cards and netbanking (launching soon), or USD by card. Credits never expire.
GST invoices
Proper tax invoices for Indian businesses, with your GSTIN; exports are zero-rated.
Transparent pricing
Every option is priced up front — resolution, duration, audio — and shown before you run.
Built for developers
Async tasks, signed webhooks, idempotency keys and per-key spend limits.
API
Two calls: create a task, get the result
Send any model's input to one endpoint. Poll the task, or pass a callbackUrl and we POST the result to you, signed.
curl -X POST https://api.anky.ai/v1/tasks \ -H "Authorization: Bearer $ANKY_API_KEY" \ -H "Content-Type: application/json" \ -d '{ "model": "gemini-3-8-flash", "input": { "prompt": "A lighthouse at sunrise, cinematic" }}'curl -X GET https://api.anky.ai/v1/tasks/task_… \ -H "Authorization: Bearer $ANKY_API_KEY"Use cases
Who it's for
Creators
Try new image and video models in the Playground without juggling subscriptions — pay only for what you generate.
Agencies
Produce client work across many models from one wallet, with per-key spend limits for each project.
Apps & products
Build features on top of the best model for each job, and switch models without rewriting your integration.
FAQ
Questions, answered
Start generating in minutes
Create an account, top up your wallet and call any model — from the Playground or your own code.