Frequently Asked Questions

How many AI models are available?

Over a thousand AI models across 11 categories. New models appear automatically as fal.ai adds them — no plugin update required. See Available Models for a category overview.

How much does it cost?

Two separate costs: a plugin subscription ($9/month or $90/year) and per-generation AI costs charged by fal.ai directly. modelBridge shows estimated cost before every generation. Full pricing details →

Do I need a fal.ai account?

Yes. modelBridge is an interface to fal.ai’s AI models. You need your own fal.ai account with an API key and billing. All generation happens on fal.ai’s infrastructure — modelBridge connects you directly.

Is my data private?

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

Can I use generated content commercially?

Commercial use rights depend on the specific model’s license, fal.ai’s terms of service, and your jurisdiction’s laws. modelBridge doesn’t add restrictions — check the model creator’s licensing terms. Use the Commercial filter chip in Model Search to find commercially licensed models.

Can I use modelBridge offline?

No. modelBridge requires an active internet connection to communicate with fal.ai’s API for generation. The plugin itself loads inside Premiere Pro without internet, but no generation or model catalog updates are possible offline.

Does it work on Windows?

Not yet. modelBridge is currently macOS only.

Which Premiere Pro version do I need?

Premiere Pro 2023 (v23.x) or later.

How accurate are cost estimates?

modelBridge uses five confidence tiers — from confirmed actuals to minimum starting prices. Estimates improve automatically as you generate: after a few runs with the same configuration, the badge levels up to a median-based “Learned” estimate. Full confidence tier details →

What happens if a generation fails?

modelBridge validates your input before sending, learns from failures to prevent repeat errors, and shows clear messages explaining what happened and what to do. Transient errors retry automatically. How errors work →

Can I keep editing while a generation runs?

Yes. Every generation runs in the background. You can switch models, launch more generations, and keep editing your timeline — you get notified when each result is ready. See Background Generations.

What is Dual Mode?

Run two models simultaneously on the same clip and compare results side by side. Useful for comparing quality, speed, voices, or generating options for clients. See Dual Mode.

Can I generate audio and voiceovers?

Yes. 10+ audio models including ElevenLabs, Kokoro, and Qwen-TTS. Generate voiceover, sound effects, or music directly in Premiere Pro — results import to the correct audio track automatically. See Audio & Voice.

Can I preview results on my phone?

Yes. Scan a QR code once, save it as an app, and every future result appears automatically. See Mobile Preview.

What happens during long generations?

They move to background processing automatically. The Active Generations Panel tracks progress while you keep editing. Polling continues for up to 30 minutes.

Can I cancel a generation?

No. Once submitted to fal.ai, a generation can’t be cancelled. You’ll be charged for completed generations. Check cost estimates and use short durations for test runs.

Can I generate without a clip on the timeline?

Yes — text-to-image, text-to-video, text-to-audio, and text-to-speech models work from text alone. You can also drag files from Finder directly onto the media card. See Drag-and-Drop.

Can I train my own models?

No. modelBridge connects to existing models on fal.ai. If you deploy a custom model on fal.ai, you can add it by entering its endpoint path.

What is Agent Mode?

An AI editing assistant built into the plugin. Describe what you want in plain language and the agent handles model selection, parameter setting, timeline edits, and QC checks. Requires your own Anthropic API key (not included in the subscription). Typical cost: ~50–60 SEK/month for regular use. See Agent Mode.

What are LM Arena rankings?

Community-driven model rankings from blind head-to-head comparisons. Updated nightly. Available for image-to-video, text-to-video, and text-to-image categories. See Trending & LM Arena Rankings.

Can I disable error reporting?

Both error telemetry and behavioral analytics are off by default. You must opt in explicitly in Settings. See Privacy Controls for what’s sent and how to control it.

How do I update the plugin?

See Updating modelBridge. All settings, history, learned constraints, and cost data persist through updates.

What happens if fal.ai has issues?

modelBridge monitors fal.ai service health automatically. If an issue is detected, you’ll see a notification directly in the plugin panel. Most generations will continue working — the plugin is designed to keep running even when parts of fal.ai’s infrastructure are degraded.

What if a model disappears?

Models are hosted by fal.ai and may be renamed, updated, or removed. Search for the model by name (it may have moved), check fal.ai’s platform, or try alternatives in the same category. Your local data for removed models is preserved in case it returns.