Pricing

Currently, the cloud version and the open-source offline version offer the same features. The main differences are:

① Using the open-source offline version requires you to deal with various installation issues on your own. The cloud version, on the other hand, is ready to use directly.

② The cloud version runs on high-performance GPUs, offering stronger performance, but at a cost. The offline open-source version uses your computer's GPU, so it's free.

If you have a low-end computer or don't want to tinker, our cloud version might be more suitable for you. You can download and install our free version (opens in a new tab) on your computer and run the default workflow to see how fast it generates images. Compare the speed with our cloud version. If the local image generation speed is faster, there is no need to use our cloud version.

No, we don't.

Firstly, as a startup, we currently lack the funds to cover the costs of offering free trials. Additionally, even if we were to offer trials, the amount of usage we could provide would be limited. For most people, it might only stretch to producing a few images, which isn't sufficient for comprehensive use.

So, if you're looking to try our product, we recommend downloading and installing our open-source version, which is perpetually free and essentially offers the same features as the paid version, especially in terms of generating images.

However, since the product is in its early stages, it might be somewhat unstable, but we're continuously making improvements. To show our appreciation for early supporters, we're offering a discount: 10% off for Hobbyist and 20% off for Pro.

Unfortunately, we do not offer a student discount at the moment. For students, we recommend downloading and installing our open-source version, which is permanently free and virtually identical to the paid version in terms of features like image generation.

Each time you run a workflow, you consume GPU credits. The number of credits consumed per run = GPU credits consumed per second x total running time of the workflow. To simplify cost calculations, the value of each credit is $0.0001, i.e., 1/10,000 of a dollar.

T4 consumes 4.8 GPU credits per second, L4 consumes 6.8 GPU credits per second, A10G consumes 7.0 GPU credits per second, A100(40G) consumes 26.1 GPU credits per second, and A100(80G) consumes 41.7 GPU credits per second.

The number of images you can generate depends on the GPU you are using and the complexity of the workflow. If you run the default workflow on an L4 GPU and use the SD1.5 model, the average time to generate an image is 30 seconds. The Hobbyist plan can generate about 700 images per month, and the Pro plan can generate over 2100 images per month. For a detailed GPU comparison, please refer to this document.

At this moment, any unused monthly GPU credit does not roll over. This is to facilitates effective capacity planning and ensure fair resource allocation among subscribers.

If you run ComfyUI on cloud servers (e.g., Google's Colab), you'll find that even just setting up a workflow consumes GPU credits. This is because these cloud servers are billed based on GPU usage time. Our billing method, on the other hand, is based on actual GPU credit usage, so you won't waste GPU credits just setting up a workflow, and you'll only consume GPU credits when generating images.

Yes, you are free to cancel your subscription at any time but the cancellation will be effective at the end of the current billing cycle. If you change your mind, you can un-cancel your plan before the end of the cycle.

Yes, you can upgrade your subscription plan at any time. The upgrade will be effective immediately and you will be charged the difference between the two plans.

Currently, we do not support the addition of custom models via the web UI. If you need a custom model, please raise a request via this form We can manually add them to the product if they are available on Civitai.

We don't support user installations of custom nodes due to the unique challenges they present and the pure serverless design of Comflowy. However, you can submit this form so that our team can manually install them.