THE CREATOR HANDBOOK / 5 MIN READ
BYOK: Bring Your Own Key
Connect your API account to create images and videos in Gen. No Gen credits are charged; your API provider bills generation costs.
01What is BYOK?
BYOK means Bring Your Own Key. You supply an API account and models; Gen provides image and video studios, task progress, previews, downloads and history.
Save multiple personal providers in one account. Before generating, choose the site channel or a personal channel. Your saved providers are private to your account.
Zero Gen credits does not mean a free API. Check your provider’s balance, model access, prices and billing rules. Gen does not quote your personal API’s charges.
02Connect your API account
- Sign in to Gen. Open Account → Media providers, then choose Add provider.
- Follow the form’s link to Nbility API. Create a key with access to your models and confirm that your API account has sufficient balance.
- Give the provider a recognizable name and paste your API key. The default Base URL is locked to https://api.nbility.ai.
- Fetch or manually add image and video models. Check their settings and save. You can also save the key first and add models later.
03Fetch models or configure them manually
Choose Fetch models. Gen requests GET /v1/models using that provider’s API key. Search the returned IDs, add a model and choose its image or video type.
Alternatively, choose Add manually and enter the exact model ID from your provider’s documentation, plus a display name. IDs are case-sensitive; do not use a marketing name or web address.
The model list primarily provides IDs, not guaranteed sizes, qualities, durations, ratios or reference support. Imported defaults are a starting point to review against provider documentation. Fetching again does not overwrite saved settings.
| Setting / question | What it means | What to do |
|---|---|---|
| Image models | Configure sizes, quality options and the reference-image limit. | Use supported sizes such as 1024x1024, auto or 1K. Add only sizes and quality values the model accepts. |
| Video models | Configure request format, duration, aspect ratio, resolution and reference support. | For standard New API models, use the size format and map each ratio to a pixel size. For multimodal models, configure resolution, ratio and content according to their documentation. |
Selecting the video type does not make an ID a video model. Both the model and its parameters must match the provider and Gen’s request format.
04Create with a personal channel
- Open Image studio or Video studio. Select your saved provider under Generation channel.
- Choose a personal model, enter a prompt and select supported settings. Keep references within the configured limits.
- Confirm the “No Gen credits charged” notice. Generate images directly; for video, review settings before confirming generation with your personal API key.
- Wait for the result, then preview or download it in the studio or History. Personal tasks identify the provider and billing source.
You can use personal channels with zero Gen credits. Switching to the site channel restores its credit checks and billing. Personal failures do not refund Gen credits; check your provider for API charges or refunds.
05Why is the Base URL locked?
By default, only https://api.nbility.ai is allowed and Base URL cannot be edited. Administrators can enable Allow custom Base URLs in Personal provider settings. It is off by default.
When enabled, only public HTTPS backends compatible with New API are supported. Enter the base URL and a key issued by that provider. Changing the URL requires re-entering a key. Chat API compatibility alone does not guarantee image or video support.
If administrators turn the switch off again, existing third-party providers show URL blocked and cannot start new tasks. Submitted tasks keep the configuration used at submission.
06Key management and troubleshooting
Keys are encrypted on Gen’s server and are not shown again when editing. Leave the field blank to keep the saved key, or enter a new one to replace it. Keep keys out of prompts, public examples and screenshots.
Editing or deleting a provider does not replace the credentials of active tasks. To revoke a key immediately, revoke it at your provider; unfinished tasks using it may then fail.
| Setting / question | What it means | What to do |
|---|---|---|
| Discovery fails / 401 or 403 | The key may be invalid, expired, lack permission, or model discovery may be unavailable. | Check the key, balance and access with your provider. If discovery is unavailable, add models manually from its documentation. Failed discovery preserves your configuration. |
| Provider missing in the studio | There may be no models for that media type, or the provider may be disabled. | Check that the provider is enabled, its URL is allowed, and it has at least one model for the studio’s media type. |
| Size, duration or reference errors | Configured capabilities may not match the model. | Correct the format and options using provider documentation, then try a simple text-only request. Avoid repeat submissions when task status is uncertain. |