Skip to content

Model Search

Model Search connects you to fal.ai’s full catalog without leaving Premiere Pro. Search by name or keyword, filter by category, and add any model with one click. It’s ready to use immediately.

Type a query — “video upscaler”, “image to video”, “kling”, “wan”, “elevenlabs” — and results appear as you type. Each result card shows:

  • Model name — first row
  • Category badge — abbreviated labels (Img-to-Vid, Txt-to-Img, Vid-to-Vid, etc.)
  • License badge — green “commercial” or amber “research” when available
  • NEW badge — orange badge for models added to fal.ai within the last 14 days
  • Key strengths — 2–3 bullets describing what the model does well, written for video editors
  • Preview — expand the card to see example output (image, video, or audio) before adding
  • Description — from the model’s API metadata

Expand any model card and you’ll see an Ask your agent section between the model’s Key Strengths and its Technical Specs. Click Explain this model and the Agent tab takes over.

The agent already knows everything the card shows you — name, category, strengths, the model’s real input parameters — so you get a short, opinionated walk-through in seconds: what the model is best for in a Premiere workflow, the parameters that actually matter for this specific model, and any gotchas worth knowing. No generic “here’s how text-to-image works” boilerplate.

When the agent offers to add the model to your project, say yes — it installs and selects it for you, ready to generate.

Six category chips sit above the results for quick filtering:

ChipWhat It Shows
AllEverything matching your query
Image GenText-to-image and image-to-image models
Video GenText-to-video and image-to-video models
UpscaleImage upscaling and enhancement
InpaintingInpainting and fill models
Extend VideoVideo extension models

A Commercial toggle works independently from category chips. Combine it with any category to show only commercially licensed models — useful when generating content for client work.

modelBridge monitors fal.ai’s catalog continuously. When a new model is published, it appears in Model Search and is announced via the news feed with an orange NEW badge. No plugin update needed.

You can also trigger a manual refresh at any time — click the refresh button at the top of Model Search to check for newly available models immediately. A notification tells you how many new models were found.

The Latest models filter shows only models published within the last 14 days, so you can quickly see what’s new without scrolling through the full catalog.

The model count above the search field always reflects the current state of fal.ai’s catalog, not an old snapshot. Every model shown has been individually confirmed to work — not just listed. The count updates automatically as fal.ai adds new models.

modelBridge is designed to stay reliable and stable: new models appear quickly, temporary platform hiccups never cause models to “disappear” by mistake, and if something is removed it’s because the platform itself changed — not because the integration glitched.

Four sort options:

  • Relevance — fal.ai’s default ranking
  • A → Z — alphabetical
  • Z → A — reverse alphabetical
  • Newest — most recently updated first

Click +Add on any search card. The plugin fetches the model’s schema and pricing data, then registers it in your collection. Visual feedback through each state:

  1. Add — default
  2. Adding… — shimmer animation while schema loads
  3. Added — green flash, card fades to indicate it’s installed

After adding, the model appears in your installed models dropdown and is ready for generation immediately.

Behind the scenes, modelBridge does five things:

  1. Fetches the OpenAPI schema from fal.ai — every input field, constraint, and default
  2. Classifies each field — determines whether it’s a slider, dropdown, text input, media upload, or checkbox
  3. Detects media requirements — what input types the model needs (image, video, audio, or none)
  4. Fetches pricing — queries fal.ai’s pricing API for cost estimation
  5. Caches everything locally — future loads are instant, no network call needed

This is what makes the full fal.ai catalog possible without any hardcoded support. Every model gets a purpose-built interface generated from its own specification.

If you know a model’s fal.ai endpoint, add it directly without searching:

  1. Use the “Add model by ID” input
  2. Enter the path (e.g., fal-ai/kling-video/v3/pro/text-to-video)
  3. The plugin fetches the schema and adds it
  • Check your internet connection — search requires access to fal.ai
  • Try a broader query — search is keyword-based, not fuzzy
  • Try a different filter chip
  • The model’s schema may be temporarily unavailable on fal.ai
  • Some models are gated and require access approval from the creator
  • Check the error banner for specific details
  • The plugin checks server health before each search — there may be a brief delay if the local server is recovering
  • Large result sets load in parallel but may take a moment on slow connections