Different products meter different operations
| Workflow | Documented unit | Budget consequence |
|---|---|---|
| HeyGen avatar models | Model-specific credits per minute | A model change can alter capacity substantially |
| Selected Arcads Talking Actor modes | 800 credits per rounded-up minute per actor | 30 and 60 seconds consume the same base amount |
| Creatify standard avatar generation | 5 credits per started 15 seconds | 31 seconds consumes more than 30 seconds |
| MakeUGC plans | Plan credits with feature-specific access | Do not convert balance into ads without the mode rate |
| Final campaign asset | Not a universal vendor billing unit | Include finishing, corrections and rejected attempts |
Rates refer only to the specified documented operations, checked September 20, 2026. Confirm the actual account offer and usage display. No cross-provider equivalence of quality or finished output is implied.
Define the avatar workload precisely
Write the number of segments, the duration of each, the number of actors and the selected model. Specify whether the input is a photo, video, audio track or another source. These details can determine the rate even inside one product.
Separate the avatar segment from the finished video. A thirty-second presenter render may still need product footage, captions, music and a closing card. Another workflow may include more of that assembly. Compare the complete deliverable rather than assuming every provider’s video count means the same thing.
Include the likely correction. A changed sentence, a new voice and a different actor can have different consequences from adjusting a caption. The budget should reflect the work you expect to perform, not an idealized first pass with no revisions.
Keep the workload small enough to verify. One representative segment and a correction can establish the relevant account behavior before you extrapolate to a monthly plan. Check the displayed usage against the documentation and ask the provider to resolve any difference.
HeyGen: model choice changes the nominal capacity
HeyGen’s current public credit guide lists Avatar IV photo at sixteen credits per minute, Avatar IV video at thirty-one and Avatar V at forty-eight. Its Creator offer lists 600 credits for $29 per month. Those figures allow a model-specific capacity calculation under stated assumptions.
Dividing 600 by the published rates gives nominal capacity of 37.5 minutes for Avatar IV photo, about 19.35 minutes for Avatar IV video and 12.5 minutes for Avatar V. This assumes all credits go to the selected operation and excludes extra features, corrections and any account-specific billing details.
The guide also lists other avatar modes and extra operations. Do not treat one calculated capacity as a universal HeyGen allowance. A workflow that uses translation, expressive motion or broader video generation needs its own usage calculation.
Use the rate shown for your actual account before purchasing or rendering. Public documentation is the basis of this comparison, but the account’s offer and usage display are the operational inputs. Resolve a discrepancy rather than silently substituting a third-party estimate.
Nominal HeyGen Creator capacity by selected model
| Model | Published credits/minute | 600 credits ÷ rate |
|---|---|---|
| Avatar IV photo | 16 | 37.5 minutes |
| Avatar IV video | 31 | About 19.35 minutes |
| Avatar V | 48 | 12.5 minutes |
A capacity allocation using the published 600-credit offer. No extras, retries or other usage; not a guarantee of finished or approved video minutes.
Arcads: rounding matters more than a simple seconds calculation
Arcads’ credit guide lists 800 credits per minute per actor for selected Talking Actor modes and rounds duration up. A single-actor thirty-second render and a sixty-second render therefore both use 800 credits under that rule. A sixty-one-second render uses 1,600.
Ten separate thirty-second renders consume 8,000 credits for this operation. Combining durations mathematically into five minutes would miss the per-render round-up behavior. Calculate each requested render using the actual rule before summing the workload.
The rate does not apply to every Arcads model or tool. Product generation, translation and other operations have separate schedules. Keep those items separate in the estimate rather than presenting a single actor rate as the cost of the whole platform.
A newer onboarding offer reported by the publisher on September 20 showed 1,000 credits, three trial days, $77 for the first month and $110/month thereafter. The calculations above describe the linked help-center rules only. We have not established that those rates use the same credit system as this newer offer, so do not apply them to its balance or infer a dollar cost per minute.
Creatify: started intervals create their own boundaries
Creatify’s usage guide lists standard avatar and video-ad generation at five credits per started fifteen seconds. A fifteen-second example consumes five credits, thirty seconds consumes ten and thirty-one seconds consumes fifteen. The boundary affects a short ad even when the duration change is small.
With the published Starter offer of $39 for 100 credits, an even subscription allocation gives $1.95 for five credits, $3.90 for ten and $5.85 for fifteen. These figures are nominal shares of the subscription, not one-off purchase prices. They assume the allowance is used and exclude other work.
Agent workflows are different. The documentation explains that usage can vary with steps and attempts, so the standard avatar formula should not be applied to every ad created through an agent. Check which operation you are actually using.
Revision conditions also matter. A finishing change and a new script, voice or avatar are not necessarily billed the same way. Test the required correction and use the current rules for the chosen plan before projecting monthly capacity.
MakeUGC: a credit balance needs a feature rate
MakeUGC’s pricing page lists recurring plans with credit balances and different feature access. Start up is shown at $59 with 500 credits, Growth at $79 with 1,000 and Pro at a recurring $149 with 2,000. Promotional and introductory offers appear alongside these plans.
Those balances do not by themselves establish a fixed number of avatar videos. Identify the selected mode, duration and any separate generation steps. If the current public offer does not resolve the cost, request a concrete workload example or verify the account’s displayed charge.
A required feature can determine the tier even when a smaller allowance would cover the nominal usage. Check product-in-hand, batch or custom capabilities under the current offer. Do not compare a plan that cannot perform the job with one that can.
Keep website and API offers separate unless the provider explicitly confirms how they relate. A developer credit package and an app subscription may have different access and billing. Use the offer that matches the way the team will produce the video.
Distinguish theoretical allocation from effective cost
Dividing a subscription price by its credits gives a nominal allocation, useful for understanding a workload’s share of capacity. It does not mean you can buy that exact number of credits at that price or that every credit will become usable output.
Effective cost depends on actual use. If a hypothetical $30 subscription produces only two approved assets in a month, the subscription component is $15 per asset, even if a fully utilized model suggests a much lower theoretical amount. This example illustrates utilization, not a named provider’s performance.
Add the cost of rejected attempts and required finishing. A segment that cannot be used still consumed production resources. Track why it was rejected so the next decision can address the cause rather than simply buying more capacity.
Keep the figures labeled. Use “calculated capacity” for documentation-based arithmetic and “observed production cost” only when you have recorded actual work. Mixing those categories can make an estimate look more certain than it is.
Account for setup, identity and team requirements
A stock avatar and a personal presenter can have different setup and access requirements. Confirm the current custom-avatar or voice entitlement before treating the base subscription as the complete cost. Additional slots or specialized modes may be separate purchasing decisions.
Team seats also need a specific check. Adding an operator does not automatically add generation capacity, and a shared allowance may behave differently from separate accounts. Read the offer rather than multiplying credits by the number of people using the tool.
If the workflow uses a separate voice service or editor, include those subscriptions and handoffs. An integrated product may reduce transfer work, but only if its finishing controls meet the brief. Compare the complete stack using the same deliverable.
Record setup effort as well as recurring work. A personal avatar may require initial preparation that becomes worthwhile over a series but is excessive for one short project. The right decision depends on the expected reuse, not just the first month’s price.
Check expiry, rollover and additional usage
Unused capacity can affect a business with an irregular production schedule. Check whether credits expire, roll over under particular conditions or reset at renewal. Do not assume a provider’s rule applies to every tier or to both monthly and annual billing.
Find out what happens at the limit. The account may stop generation, offer a top-up or require a plan change. That behavior matters when the team needs a small correction near a deadline.
Keep annual allocation separate from annual payment. A yearly subscription may still release credits periodically. Paying for twelve months does not necessarily make the entire year’s capacity available at once.
Save the current offer and relevant usage terms with the purchasing record. If pricing changes later, the business should know which plan and conditions it used for the original estimate. This is especially useful when comparing a legacy account with a newly advertised plan.
Build a model-specific purchasing worksheet
For each candidate, record the subscription amount, billing period, allowance, exact mode, duration rule and required extras. Mark missing inputs visibly. A blank field is more honest and useful than an invented average rate.
Calculate the planned first pass, then add a separately stated correction allowance based on your pilot. Do not hide retries inside a vague cost-per-video claim. Keeping them separate shows which assumptions drive the estimate.
Check that the planned output is actually accepted. A low-cost render with wrong pronunciation or unusable product details does not satisfy the brief. The purchasing decision should combine usage evidence with editorial acceptance.
Review the estimate after one normal production cycle. Compare planned and actual usage, unused capacity and finishing work. Adjust the plan or process when the evidence supports it, rather than repeatedly buying more credits without understanding where they go.
Avatar pricing worksheet
| Input | Example question |
|---|---|
| Offer | What is the actual recurring charge and billing period? |
| Mode | Which exact avatar model or generation operation? |
| Usage rule | Per minute, rounded interval, actor or variable agent steps? |
| Extras | Voice, translation, motion, editing or additional seats? |
| Correction | What does the expected revision consume? |
| Acceptance | How many outputs meet the complete brief? |
A proposed comparison worksheet. Missing rates should remain unknown until verified.
Common avatar pricing questions
Can you compare cost per minute? Yes, for a defined model and workload with a documented rate, but label the assumptions. A universal number across all modes hides important differences in generation and finishing.
Are shorter videos always cheaper? Not under every rule. Arcads’ selected Talking Actor modes round to minutes, while Creatify’s standard avatar mode uses started fifteen-second intervals. Check the actual duration boundary.
Is a larger credit balance better value? Only after applying the feature rate and confirming that you will use the capacity. Credits from different providers are not interchangeable.
Which tool has the lowest cost per approved video? We have not measured that across tools. Use the same representative brief, record actual attempts and finishing, then calculate the accepted-output cost for your workflow.
Watch the duration boundary during script revisions
A correction can move a segment across a billing interval even when the message changes only slightly. Keep the planned and final duration visible in the estimate. Under the selected Arcads rule, moving from sixty to sixty-one seconds changes the rounded actor-minute count; under Creatify’s standard rule, moving from thirty to thirty-one seconds changes the fifteen-second interval count.
Do not shorten a necessary qualification solely to stay below a boundary. First check whether repetition can be removed without changing the meaning. If the longer version is needed, budget for it. The correct purchasing comparison is the cost of the accurate deliverable.
When a batch contains different lengths, calculate them individually before summing. An average duration can hide how many segments cross a threshold. Keep the per-segment record with the workload so another person can verify the arithmetic and update it when the script changes.
This is also why a single advertised cost per video is rarely enough. Duration rules, selected modes and revisions all determine which production operations the subscription must cover.
What we would do
Price the exact avatar model and production route. Apply the documented duration rule, keep extras and corrections visible, and distinguish theoretical capacity from actual accepted output. Confirm the account offer before turning an estimate into a purchase.
Sources & methodology
We reviewed public vendor pages and search results on . We did not run paid hands-on tests or measure ad performance. “Verified” means supported by a linked official source, not independently tested. “Calculated” means arithmetic with stated assumptions. “Reported” means information supplied by the publisher or a third party, with its provenance stated in the article. “Unknown” means not established in this review. “Editorial” identifies our analysis.
- HeyGen: official pricing information ↗
- Arcads: official product information ↗
- Creatify: official pricing information ↗
- MakeUGC: official pricing information ↗
- HeyGen: avatar and feature credit rates ↗
- HeyGen: current subscriptions ↗
- Arcads: model-specific credit rules ↗
- Creatify: standard and agent usage ↗
- Creatify: current subscriptions ↗
- MakeUGC: current offers and features ↗
Prices are in USD where shown. Confirm billing period, applicable tax, promotions, feature access and usage rules at checkout. Research dates are fixed to this review, not automatically refreshed on deployment.