Google AI Studio

Google AI Studio (/media/google-ai) exposes Google's Gemini Developer API for media generation — Nano Banana and Imagen image models plus Veo video — using a single Gemini API key. It is the consumer/developer path, distinct from the enterprise GCP path at Google Vertex.

Where to configure

This is a universal-credential provider: it reuses the same Gemini API key you configure under Settings → AI → Google Gemini. No separate media credential is required. See Settings for AI provider setup.

Tabs

Tab	Operation	Key fields	Produces
Text → Image	Image	Model (Nano Banana / Imagen 3/4 variants), prompt, aspect ratio, sample count (Imagen only)	Still image
Text → Video	Video	Veo model, prompt, negative prompt, aspect ratio, resolution, duration	MP4 video clip

Generation flow

Generations follow the standard Media Studios flow: the job enters the Render Queue, completes, and lands in /files. Finished images and videos can be opened in the Designer or posted straight to the composer.

Caveats

• Image is synchronous. Nano Banana and Imagen complete inline and appear almost immediately.
• Video is asynchronous. Veo uses :predictLongRunning and completes via the media-jobs-poll cron — Google provides no completion webhook.
• Auth-only Veo download. The finished MP4 is downloaded with the key and returned as a data URL so the lifecycle can import it without an unauthenticated re-download failing with 401.
• One Gemini key drives both the LLM and media surfaces; configure it once.

---

Verified against v1.0.0