Moving from Processing Hours to Processing Credits

We’ve made a small change to how usage is measured in Invoke. Instead of “processing hours,” you’ll now see “processing credits” in your account. This update doesn’t change how much you can do with your plan—it’s simply a more flexible system that enables us to support a wider range of models going forward.

What are processing credits?

Credits are the new unit we use to measure compute usage across all models in the platform.

• 1 credit = 1 second of compute time on the models we host ourselves.
• Your plan’s usage hasn’t changed—just the unit of measurement.

For example, if your plan previously included 5 hours of processing time per month, that now becomes:

5 hours x 60 minutes x 60 seconds = 18,000 credits

When are Processing Credits used?

Processing credits are calculated based on the actual compute required to generate images and run models using our resources.  

This includes any task where our system is actively creating / processing data or requesting an external API, such as:  

• Generating images based on your input.
• Running Workflows that generate images.
• Running model training jobs.

Processing credits are not being charged for tasks that do not require substantial computational resources. These include:

• Designing, editing, or setting up workflows.
• Uploading or managing files and assets.
• Planning or configuring settings.
• Browsing through the application or idle time within the studio space.

Our goal is to ensure that you are only billed for the resources you actually use for generating outputs, ensuring efficient and cost-effective usage of compute.

Why switch from hours to credits?

We’re expanding the platform to include new models that offer enhanced capabilities, but some of those models are accessed differently under the hood.

Currently, models in Invoke run on our own infrastructure. That means we can track how long each generation takes, and deduct usage accordingly—1 second = 1 credit.

However, some models are accessed via API, which simply means we’re securely sending your request to another service and receiving the result back. These external providers don’t charge by the second—they charge per request. Because of this, time-based metering isn’t possible for those models.

Moving to a credit system lets us support both:

Which models will use API-based credits?

We’re actively working on support for new models, including:

• Imagen 3 – a high-fidelity image generation model
• ChatGPT 4o – OpenAI’s advanced multimodal model, for creative or text+image workflows

As these become available, we’ll share more details about how they work and how many credits they require per use. They’ll be fully integrated into the Invoke interface—no need to manage separate tools or accounts.

Does anything change in how I use the Invoke application?

No—everything works the same as it does today. You’ll continue to create and edit images using the same workflows. You’ll just see credits listed instead of processing hours, and in the future, you may see new models added to the model selector with clearly labeled credit costs per use.

Will my plan include fewer resources?

Not at all. The credit amounts assigned to each plan are directly equivalent to the compute time previously included.

Still have questions?

Feel free to contact support or check the release notes for updates as we roll out new models.