The key idea
Gammal Tech includes 200 signed-in AI calls per project each month. Extra signed-in calls cost 0.20 EGP each. Guest calls cost 0.40 EGP and the first guest call consumes the remaining free allowance.
What does an AI API actually charge for?
AI API pricing becomes easier to compare when you identify the billing unit. A call is one request to the AI service. Tokens are pieces of text used to measure input and output. A conversation may contain several calls, and each call may contain a different amount of text.
For a website assistant, a visitor asking three questions will normally trigger three AI requests if each question is sent separately. Sending a second request to rewrite an answer adds another call. A quota of 200 calls therefore does not mean 200 visitors or 200 complete conversations.
Start by drawing the actual interaction: visitor question, AI request, displayed answer, and possible follow-up. This reveals the usage you need to estimate before comparing headline prices.
How to make a fair AI API pricing comparison
Compare services against the same small collection of website tasks. For example, use a delivery-policy question, a product explanation, and a request to simplify a paragraph. Specify roughly how much context you send and how long the response should be.
| Pricing unit | What to measure | What can change the estimate |
|---|---|---|
| Per call | Total AI requests | Follow-ups, retries and extra processing steps |
| Per token | Input and output text volume | Conversation history, context length and response length |
| Subscription or package | Included usage and overage rules | Plan limits, renewal terms and paid extras |
A low unit price alone cannot establish the cheapest option. Check website support, authentication requirements, usage limits, and whether the answers are suitable for your task. For a Saudi business comparing budgets in riyals, keep the provider's original currency visible and apply a current exchange rate separately.
Gammal Tech's allowance and prices
Gammal Tech's website AI service uses call-based pricing. Each project receives 200 free AI calls per month for signed-in users. After those calls are used, signed-in AI calls cost 0.20 EGP each. The allowance resets monthly.
Guest AI calls have no free allowance and cost 0.40 EGP each. There is an important interaction between these modes: the first guest AI call also consumes all remaining signed-in free calls for that project for the month. Guest access is therefore a budget decision, even if only one guest uses it.
The first project is free to create. Additional projects have a one-time fee, whose amount should be checked before creation. Keep that project cost separate from recurring AI usage.
Estimate a month of signed-in usage
If a project uses only signed-in AI calls, subtract its 200-call monthly allowance from the month's total and multiply any remaining calls by 0.20 EGP. These hypothetical examples describe AI call charges only.
| Signed-in calls | Paid calls | AI usage cost |
|---|---|---|
| 150 | 0 | 0 EGP |
| 300 | 100 | 20 EGP |
| 700 | 500 | 100 EGP |
Estimate requests from expected interactions, not page views. For example, 40 signed-in visitors asking three questions each would generate 120 calls if the website makes exactly one request per question. That is a planning assumption, not a prediction of visitor behaviour.
Calculate the effect of enabling guest AI
Suppose a project starts the month with 50 signed-in calls. Those calls use part of its allowance, leaving 150 free calls. One guest then sends an AI request: that request costs 0.40 EGP, and the remaining 150 free signed-in calls are consumed.
If signed-in users subsequently make another 50 calls that month, those calls cost 10 EGP. Total AI charges in this example are 10.40 EGP. Without the guest request, the 100 signed-in calls would have remained within the free allowance.
A separate example is 50 guest calls with no other AI usage, costing 20 EGP. The free signed-in allowance does not reduce that guest bill. Decide which experience you want before exposing a guest assistant on the website.
Reduce unnecessary requests before increasing the budget
Show useful existing help content before inviting an AI question. Prevent duplicate submissions while a request is pending. Avoid automatically making extra AI requests to rephrase every answer unless that step solves a demonstrated problem.
Supply concise, current business facts so visitors can receive a useful answer with fewer clarification turns. Test the conversation with representative questions, then record actual request counts to replace your early assumptions.
Gammal Tech uses whitelisted website domains rather than API keys. Keep integration choices within its website scope, and remember that a project identifier is public, not a secret credential. Cost planning should include the user experience needed for visitors to sign in.
Common questions about AI API costs
Are 200 free calls enough for a production website?
They can support a small workload or an initial trial, but adequacy depends on actual requests. Measure questions per visitor and any additional calls your interface makes before choosing a budget.
Can I offer free AI to every guest?
The project's free allowance applies to signed-in users. Guest calls are charged from the first call, which also consumes the project's remaining free signed-in allowance for that month.
Should I compare only the cost per call?
No. Compare the complete task, including answer usefulness, the number of requests required, website compatibility, and authentication. A call price and a token price need a shared workload before they can support a meaningful comparison.
Source and service details
Integration and pricing details are based on the Gammal Tech guide. Check the documentation before publishing or changing your project configuration.