You now have full control over the quality, speed, and cost of every generation, across both the web app and the API.
Resolution
Choose between three resolution tiers for every generation:
- Auto (default): Matches output resolution to your input image.
- 1K (~1 megapixel): Fast, consistent results with strong instruction-following.
- 2K (~4 megapixels): A step up in detail while keeping generation times reasonable.
- 4K (~16 megapixels): Maximum detail for high-fidelity output.
Generation Mode
A new parameter that lets you decide how to balance speed, cost, and quality:
- Auto (default): Selects the optimal mode based on your resolution, tool, and plan.
- Fast: Quickest turnaround at the lowest cost.
- Balanced: A middle ground between speed and realism.
- Quality: The most detailed and realistic output, with longer processing times.
Generation Mode is available on all image generation tools. Try-On currently supports Balanced and Quality only.
Credit Costs
Credit costs now scale with the resolution and generation mode you choose, so you only pay more when you need more.
What This Means for App Users
Most configurations are now cheaper or the same. Fast mode at any resolution costs just 1 credit. The only configuration that costs more is Quality + 4K at 3 credits. We've also reduced the Face Reference surcharge from 3 credits to 2 credits per image. See full breakdown in App Credit Costs.
What This Means for API Users
Fast + 1K behaves the same as before at the same cost. Higher resolution tiers produce more realistic output and are priced accordingly. Try-On Max can now run at 2 credits with Balanced + 1K (previously 4 credits) with similar quality and faster processing. We're working on enabling Fast mode for Try-On Max so you'll be able to run high-quality 1K try-ons for just 1 credit. See full breakdown in API Credit Costs.
For a deep dive into how these settings work, check out our Resolution & Generation Mode guide.