Feature request: bring-your-own API keys for image and video generation in AI Studio
Summary
Let users connect their own provider API keys for image and video generation, as an alternative to AI Studio credits.
Problem
Many studios already pay providers directly. With credits as the only route, they pay twice for the same models and can't use models they already have access to. That pushes generation work out of AI Studio and into other tools.
Requested behavior
1. A provider panel in Settings for adding and testing API keys (Google Gemini / Imagen / Veo, Kling, ByteDance Seedance, Runway, OpenAI, fal and similar).
2. A choice on each job: AI Studio credits or a user key.
3. Keys stored locally and encrypted, never synced or logged.
4. Per-provider usage and errors shown in the job panel, including the provider's actual error message.
5. Optional: a custom endpoint for a local ComfyUI server (for example
http://127.0.0.1:8188), so users can generate on their own GPU.
Acceptance criteria
- A user can add a key, test it and run an image or video job without spending AI Studio credits.
- If a key fails, the user sees the provider's error and can fall back to credits.
- Model parameters (resolution, duration, seed) pass through to the provider unchanged.
Environment
Windows 11, RTX 5090, AI Studio with iClone 8 and Character Creator 5.
OSFC