Settings

Generation models

Choose the model for prompt, image, and video generation. Grayed-out options are placeholders until their env keys and wiring are in place.

Prompt modelModel id: gemini-3.6-flashImage modelModel id: gemini-3.1-flash-image-previewVideo modelModel id: veo-3.1-generate-preview

Live web search— grounds Trends, Dupe & Sizing in real-time results.

Gemini and OpenAI models are wired to live generation (given their API keys). Video is currently a preview placeholder.

Integration status

Live at request time. Shows whether each secret is present — values are never displayed.

ConfiguredDATABASE_URL
Persists products, generations, and creations. Falls back to in-memory storage when unset.
ConfiguredBLOB_READ_WRITE_TOKEN
Saves uploads and AI renders to Vercel Blob. Without it, assets become placeholders / inline data URLs.
ConfiguredGEMINI_API_KEY
Powers prompt building, product image analysis, and sketch visualization. Watch for 429 (quota) even when set.
ConfiguredOPENAI_API_KEY
Required when the prompt or image model is set to OpenAI (GPT-5.6 Sol / GPT Image 2). Only needed if you select the OpenAI provider.

Configure secrets in Vercel project settings, then redeploy. For public image URLs, connect a Vercel Blob store to the project so a read-write BLOB_READ_WRITE_TOKEN is injected.