Credits turn very different jobs, such as an image, a minute of voice or a video clip, into one balance. Plans usually include a monthly allowance, with top-up packs for heavy use. Rules on rollover, expiry and which features use credits vary by tool.
How it works
Behind every AI generation is compute: time on expensive GPUs running the model. Providers price that compute in credits rather than money so they can charge different amounts for different jobs without changing the plan price. The credit cost of a task usually depends on things that drive compute. For video, that is length, resolution and model quality, so a longer, higher-resolution clip costs more. For images, it is size, number of variations and the model used. For voice, it is usually characters or minutes of audio. For text, credits are often a wrapper around tokens. Your plan gives a monthly allowance that resets on the billing date. Some tools let unused credits roll over, often with a cap; others expire them. Top-up credits may have different expiry rules from plan credits. Failed or rejected generations may or may not be refunded. Because each tool defines its own credit, 100 credits in one app is not comparable to 100 in another; compare by cost per finished output instead.
Why it matters
Credits decide what an AI tool really costs you. A cheap plan can become expensive if your workflow needs many retries or high-resolution exports. For creators, knowing the credit cost of a typical video or voiceover lets you budget a month of content. For businesses, credits affect how many staff can share an account and whether usage is predictable. Reading the credit rules before buying helps avoid running out mid-project or paying for allowances you never use.
Where you will see it
- Budgeting how many short videos a month a video plan supports
- Comparing voice tools by cost per finished minute of audio
- Choosing a lower resolution draft before spending credits on a final render
- Sharing a team credit pool across several editors
Limits and common misconceptions
- Credits are not standardised, so comparing tools by credit count alone is misleading.
- Retries and failed generations can use credits, so real costs are often higher than the sticker estimate.
- Plans that expire unused credits can mean paying for capacity you never use.
How it differs from related terms
- AI credits vs Token Tokens measure how much text a language model processes; credits are a billing unit a tool sets, which may be based on tokens or on something else entirely.
Tools that use it
Frequently asked questions
Do unused AI credits roll over?
It depends on the tool. Some roll unused credits into the next month, often up to a cap, while others reset to zero on each billing date. Purchased top-up packs sometimes last longer than monthly plan credits. Check the plan page or help centre before relying on rollover.
Why do some AI generations cost more credits than others?
Credit costs track the computing work involved. Longer videos, higher resolutions, more image variations, premium models and longer audio all need more GPU time, so they use more credits. Simple tasks like a short text answer or a low-resolution preview usually cost much less.
Related terms
- Zero-shot Asking an AI model to perform a task from instructions alone, without giving it any worked examples of the desired output.
- AI agent An AI system that pursues a goal over multiple steps, choosing actions and using tools such as browsers, apps or code without step-by-step instructions.
- Algorithm A precise, finite sequence of steps that turns an input into an output, such as sorting a list or ranking search results.
- API An application programming interface: a defined set of requests and responses that lets one program use another program's data or features.
- Artificial general intelligence (AGI) A hypothetical AI able to learn and perform nearly any intellectual task a human can, adapting to new problems without task-specific training.