Skip to content

Frequently Asked Questions

modelBridge.ai has two separate costs:

  1. Plugin subscription: $7/month for access to the plugin
  2. AI generation costs: Pay-per-use, charged by fal.ai directly to your fal.ai account

Generation costs vary by model. A text-to-image generation might cost $0.01-0.05, while a high-quality video generation could cost $0.20-3.00 or more depending on the model, duration, and resolution.

Yes. modelBridge.ai is an interface to fal.ai’s AI models. You need your own fal.ai account with an API key and billing set up. The plugin connects directly to fal.ai — all generation happens on their infrastructure.

modelBridge.ai provides access to 900+ generative AI models across 11 categories: text-to-video, image-to-video, video-to-video, text-to-image, image-to-image, inpainting, text-to-audio, text-to-speech, audio-to-audio, audio-to-video, and speech-to-speech. New models appear automatically as fal.ai adds them — no plugin update required.

Yes. Your API key, generation history, prompts, and settings are stored locally on your machine. modelBridge.ai does not collect analytics, track usage, or send your data to its own servers. Media files are uploaded directly from your machine to fal.ai for processing. See the full Privacy Policy for details.

Can I use AI-generated content commercially?

Section titled “Can I use AI-generated content commercially?”

The commercial use rights for AI-generated content depend on:

  • The specific AI model’s license terms
  • fal.ai’s terms of service
  • Your jurisdiction’s laws regarding AI-generated content

modelBridge.ai does not impose additional restrictions on how you use generated content. Check the model creator’s licensing terms for specifics.

No. modelBridge.ai is currently macOS only. The backend server relies on macOS-specific tools. There is no Windows version at this time.

Adobe Premiere Pro 2023 (version 23.x) or later. The extension uses CEP 11, which is available in Premiere Pro 2023 and all subsequent versions.

Cost accuracy depends on the model:

  • “Estimated” badge — typically within a few cents of the actual cost, based on curated per-model pricing data that accounts for duration, resolution, and audio settings
  • “From $X.XX” badge — shows the minimum base rate from fal.ai’s pricing API; actual cost may be higher
  • “Price not available” — no estimate is possible; check your fal.ai dashboard after generating

After generation, the Billing tab shows the verified actual cost based on fal.ai’s billing data when available.

modelBridge.ai has a three-layer error handling system:

  1. Prevention — validates your input before sending it to the API (dimensions, duration, file size, required fields)
  2. Self-learning — when a generation fails due to an undeclared model constraint, the plugin learns that constraint and enforces it automatically on future attempts
  3. Clear error messages — every error follows the pattern: what happened, why, and what to do next

For transient errors (timeouts, server load), the plugin retries automatically. For input errors, you get specific guidance on what to fix.

No. modelBridge.ai connects to existing models on fal.ai’s platform. Custom model training must be done through fal.ai directly or other training platforms. Once a custom model is deployed on fal.ai, you can add it to modelBridge.ai by entering its endpoint path.

See the Updating modelBridge guide for step-by-step instructions. All your settings, generation history, learned constraints, and cost data persist through updates.

Models are hosted by fal.ai and may be renamed, updated, or removed. If a model stops working:

  1. Search for it by name — it may have moved to a new endpoint
  2. Check fal.ai’s platform for the model’s current status
  3. Try alternative models in the same category

modelBridge.ai’s self-learning data and generation history for removed models are preserved locally in case the model returns.

Dual Mode lets you run two AI models simultaneously and compare results side by side. Select a primary and secondary model, click Generate, and both results appear as selectable cards. See the Dual Mode page for details.

Yes. modelBridge supports 10+ audio models including ElevenLabs, Kokoro, and Qwen-TTS. Generate voiceover, sound effects, or music directly inside Premiere Pro with automatic audio track import. See Audio & Voice Generation.

Yes. Mobile Preview sends results to your phone via QR code as a lightweight Progressive Web App. See Mobile Preview.

When a generation takes longer than expected, it automatically moves to background processing. The Active Generations Panel tracks progress while you continue editing. Background polling continues for up to 30 minutes.

Can I cancel a generation once it has started?

Section titled “Can I cancel a generation once it has started?”

No. Once a generation is submitted to fal.ai, it cannot be cancelled mid-flight. The generation will complete and the result will appear in the Active Generations Panel if it moves to the background. You will be charged by fal.ai for completed generations.

To minimize unwanted charges, check the cost estimate and parameters carefully before clicking Generate. Use short durations and lower resolutions for test runs.

Can I use modelBridge without a clip on the timeline?

Section titled “Can I use modelBridge without a clip on the timeline?”

Yes, for certain model types:

  • Text-to-image and text-to-video models generate from a text prompt alone — no source media needed
  • Text-to-audio and text-to-speech models also require only text input

Most image-to-video, video-to-video, and inpainting models require a source clip or image. The media card will tell you whether the selected model needs media input.

You can also drag a file from Finder directly onto the media card without importing it into Premiere Pro first. See Drag-and-Drop.

Can I generate from a file outside my Premiere Pro project?

Section titled “Can I generate from a file outside my Premiere Pro project?”

Yes. Drag an image or video file from Finder directly onto the media card — no Premiere Pro import needed. See Drag-and-Drop.