The Graydient AI rendering API
If you are building an app, an internal tool or a render pipeline, Graydient gives you open image, video and LLM models on dedicated GPUs, behind one API, at a flat monthly price.
Which plan includes it
The personal API comes with Video Max only: $84 per month on yearly billing ($999 billed yearly) or $99.99 monthly. Images and Video Plus do not include API access. You can upgrade, downgrade or cancel anytime in the billing portal, with prorated charges.
The personal API is a key for one user, and it is not for resale. It is rate-limited: keep to about one call every 10 seconds and avoid batching. Keys can expire automatically if too many requests are sent.
Building a multi-user or white-label product
Virtual users and white-label login are business-level features for deployed apps, meaning products with many users. They work on the personal API, so you can build and test multi-user and white-label flows on Video Max during development.
Going live with a multi-user product, or reselling access, needs the business-level option: a dedicated GPU server or a business agreement. To discuss a business deployment, write to info@graydient.ai.
| Personal API (Video Max) | Business deployment | |
|---|---|---|
| Users | One user (you) | Your product's users |
| Resale | Not allowed | Allowed |
| Virtual users and white-label login | Build and test | Go live |
| How to get it | Subscribe to Video Max | Dedicated GPU server or business agreement |
How the API works
The current version is V3, at https://cloud.graydient.ai/api/v3/. It follows the JSON:API specification: send and accept application/vnd.api+json, authenticate with a standard HTTP Bearer token from your API tokens page, and read results from a data object with an attributes member.
Rendering is asynchronous. You POST to POST /api/v3/render/ with a prompt and a required callback_url. Graydient replies right away with a render_hash plus estimated wait and render times. When the job finishes, it POSTs the results to your webhook. Optional fields include:
- error_url: a webhook for failures
- init_image and mask_image: for image-to-image and inpainting
- placeholders: named image URLs for ControlNet modes or IP adapters
- progressive_return: one callback per image
- session_id: correlate renders with your own users
Workflows run through the same endpoint with a workflow command in the prompt.
Status without polling: GET GET /api/v3/render/{render_hash} returns a render's status, but the docs ask you not to poll it. Use the webhook, or the WebSocket stream at wss://cloud.graydient.ai/render-events/{render_hash}, which sends started, image_done, done and error events.
Other endpoint families
- Chat: LLM chat with a persona. Replies go to a callback or come back inline with sync: true, and vision models accept an image_url.
- Concepts: list available models and LoRAs.
- Workflows: list workflows and what each supports, such as txt2img, img2img, txt2vid and img2vid.
- Knowledge base: create a domain, add facts, crawl URLs, then search or chat.
- Skills: turn natural-language requests into workflow commands.
- Virtual users: map your end users to records on your account. Build and test with it on the personal API, then go live on a business deployment.
Full reference: API documentation. The docs say they are still developing.
Typical use cases
- Powering your own app: use Graydient as the render backend for an image, video or chat feature. Develop on the personal API, then move to a business deployment when real users arrive.
- Using your hosted models: render with the LoRAs and checkpoints you have installed. See host your own AI models.
- Running ComfyAPI-style workflows remotely: call mapped workflows from your own code.
- Product and catalog imagery: upscale, edit or restyle images from a pipeline, within the personal API's rate limit.
- Batch rendering: for sustained, high-volume batches, use a dedicated GPU server instead of the personal API.
- Knowledge-base chat: build a support or documentation assistant on the knowledge base and chat endpoints.
Dedicated GPU servers
If you plan to run a multi-user product, resell API access or batch renders all day, Graydient offers a dedicated Video Max GPU server at $499.99 per month. You get your own machine with no queue, you can run longer jobs and set limits on what your customers can do and how long their queue slots run. You can stack several servers as you grow, and the contract asks for 30 days' notice, with no long-term commitment.
Render pricing on dedicated servers: renders on a dedicated server are $0.02 each, video included. This per-render price applies only to the dedicated server. The Images, Video Plus and Video Max plans stay flat-rate, with no per-render charges.
To rent one, write to info@graydient.ai.
Getting started
- Subscribe to Video Max. You can change or cancel your plan anytime in the billing portal.
- Create a token on your API tokens page.
- Read the API documentation and look on GitHub for example uses of the Graydient API.
- Set up a webhook endpoint, then send your first render request.
- Mirror and thumbnail result images before you serve them to your users, as the docs recommend.
API support is available by email at info@graydient.ai. All use is subject to the acceptable use policy and the terms of use.