How much is the ChatLLM subscription?

A ChatLLM subscription is $10 per user per month. You will be billed $10 when you sign up. You can cancel anytime, and your subscription will stay active for the month.

You can also upgrade to the Pro tier for an additional $10 a month. The Pro tier gives you more unrestricted access to our more powerful DeepAgent.  You can use DeepAgent for however many tasks you like, as long as you have active credits in the account.

What is the Pro tier and what advantages does it give me?

The Pro tier costs an additional $10 monthly and gives you unrestricted access to DeepAgent. You can use all of your active credits on DeepAgent tasks.

The Pro tier also uses a much more powerful version of DeepAgent. This means your tasks will yield better and richer results.

Finally, the Pro Tier gives you an additional 5000 credits. Please read about credits below.

What happened to Compute Points? Why are you introducing credits?

We are switching from compute points to credits for better clarity and simplicity. There is NO CHANGE to the amount of credits or compute points allocated, we are just using a different denomination.

100 compute points = 1 credit. This means that instead of 2,000,000 (2M) compute points, you get 20,000 (20K) credits. You can read more about how 20K credits translate to actual usage in the FAQ below.

What are the usage limits on the LLMs?

Our LLM usage limits are more generous than other paid services. On SOTA LLMs like GPT-4.1, GPT-4o, Sonnet-3.7, or Sonnet-3.5, you can send thousands of messages without attachments without hitting a usage limit.

We give you 20K credits every month. These credits can be used for various operations, including sending messages to the top LLMs, creating videos and images, and generating documents.

PLEASE NOTE - these credits are NOT TOKENS, and you can get up to 15M input tokens for some expensive models like Sonnet-3.7 and GPT-4.1 with these credits and 70M tokens on others.

In addition, you can use the GPT-4.1-Mini, Gemini 2.0 Flash, DeepSeek R1, and Llama 4 models without ANY LIMITS.

For coding tasks, CodeLLM provides access to specialized models, including Qwen 3, Sonnet 3.7, Sonnet 3.5, and others. These models are optimized for code generation, debugging, and refactoring.

If you choose to use credits for video, audio, or document generation, we will deduct credits proportional to the current market rate of those services. Typically, this translates to around 500 images, 40 5-second videos, or 25 DeepAgent tasks in a month from a premium service. KlingAI is extremely expensive; you will get very few videos if you use this service.

Our SOTA image models include GPT Image, FLUX-1 PRO, Recraft, Ideogram, and Dall-e. For high-quality image generation, we recommend FLUX-1 PRO and GPT Image. For video generation, our SOTA video models include KlingAI, Lumalabs, Hailuo, and RunwayML. These models are compute-intensive and may consume more credits per task.

What models can I use on an unlimited basis?

You can use the GPT-4.1-Mini, Gemini 2.0 Flash, DeepSeek R1, Qwen, and Llama 4 models on an unlimited basis.

Typically, any new model from the Llama, DeepSeek, Gemini Flash, Qwen, and GPT-4.1 Mini families will also be offered on an unlimited basis.

It’s important to note that these will still accrue credits on an ongoing basis, but you will be able to continue using them even after hitting the credit limit.

Where can I see the credits?

You can always see credits on your profile's billing page. For a detailed log on your usage, click 'View Credit Usage' on your profile. We also show credits against each message if you are using them up quickly. If you want to always view the credits used for each message, you can do so by activating the toggle on your profile.

Do my purchased credits disappear if I don’t use them within the month?

No, your purchased credits roll over to the following month if unused.

Please note that this is separate from the monthly 20K credits that come with your subscription – these reset to 20K on your renewal date.

Why don't you have exact credits published per model?

Credits are just a measure of the usage of our entire service. This includes our AI engineer, AI agents, LLMs, Video Generators, and Image Generators. We constantly add new models, change model rates, and make the product more efficient. This is why you can only see credits after usage. That said, we are very generous concerning our rate limits, and you can always monitor your credits on the billing page. We will show it for every generation or message when we think you are using too many.

Some models, such as o3 and GPT-4.5, are more compute-intensive and may consume more credits per task. This is why we recommend monitoring your credit usage on the billing page.

What is this 75% usage message I’m receiving?

The limit exists because the 10K credits cost $10, and we give you a generous amount of 20K credits every month, but the idea is to spread it for the entire month. If you want to spend more credits, you can always buy more.

You can use the last 25% on the last week of your subscription

How many users can I invite?

There is no limit to the number of users you can invite! You will be billed on a monthly basis based on the size of your team. In addition, you will be billed immediately for every additional user you invite at a rate of $10/user/month.

If I get invited to a team, do I have to pay?

No, the owner of the team organization will get the invoice for your membership on that team. If you are being prompted to pay, you likely need to switch organizations so you are on the team organization and not your own organization.

How can I view my past payments?

You can view all past payments by going to your profile and navigating to the Billing and Invoices page. There, you will be able to see your payment details and history. You will also be able to download past invoices and update your payment information.

How can I change my billing/invoice information?

Click on your Profile at the top right, then click on Profile & Billing. Click on Billing Details, then Update Address. Similarly, on the Billing Details page, you can also update your Card information.

How can I cancel my subscription?

You can cancel anytime from the Abacus.AI ChatLLM interface. Billing subscription details are available on the desktop interface by clicking on your name or organization in the top right corner and on Profile. “Cancel Subscription” should be located at the bottom right of the page.

How do I delete my credit card from my account?

To delete your credit card, you need to delete your account—email support@abacus.ai for assistance.

How do I delete my account?

In order to delete your account, please contact us at support@abacus.ai and we will handle it for you.

I would like to access certain Enterprise features. What is the price point of the enterprise plan?

The starting price point for Abacus.AI Enterprise is $5000. If you are interested, email support@abacus.ai or sales@abacus.ai to be properly connected to the right person.

There’s been a mistake in my billing. How do I get this fixed?

If you think your billing is incorrect, email support@abacus.ai and someone will investigate.

What are the limitations for the commercial use of images generated using the Abacus.AI platform?

There are no restrictions or limitations. No one owns the rights of AI generated images, so you can use them freely across any sector.

My AI model or agent isn't doing exactly what I want. Can I get a refund?

Our AI models and agents are provided as-is. Their behavior is influenced by how they are prompted, the nature of the tasks you're asking them to perform, and the complexity involved. While we continuously work to enhance their capabilities, we can't guarantee perfect performance for every prompt or scenario.

We recommend that you simplify or clarify your instructions. You can check out our Prompting Tips and Tricks. Even small adjustments to your prompt or task complexity can lead to significantly better outcomes.

Keep in mind that some tasks may be too complex for an AI—or even a human—to complete in just a few minutes.

As with most AI services, we do not offer refunds based on the model's output or performance.