Credits
Promptix is prepaid. You buy credits in Tunisian dinars and every API request is deducted from your balance. 1 credit = 1 TND. Your balance is shown in the dashboard header and on the Billing page, with three decimals (millimes).
- All API keys of your account share the same balance.
- Playground messages in the dashboard are billed exactly like API requests.
- Requests need a balance of at least 0.100 TND to start. See Insufficient credits.
- You receive an email when your balance drops below 5 TND (you can turn it off in Settings).
Model prices
Each model has two prices, in TND per 1 million tokens: one for input tokens (your prompt) and one for output tokens (the answer). They are listed in the model catalog and on the pricing page, which has a cost calculator. There is no subscription and no per-request fee: you only pay for the tokens you use.
The catalog price gives the expected cost of a request. The exact amount deducted is computed from the usage reported by the provider for that request (which can reflect cached-prompt discounts or reasoning tokens), and you can look it up afterwards, as shown below.
Topping up
Top up from Dashboard → Billing. Choose the amount of credits (from 5 to 1,000 TND per payment) and pay with Flouci. Credits are added to your balance as soon as the payment is confirmed.
Two charges are added to the amount of credits you buy, and the full breakdown is shown before you pay:
- Tunisian VAT at 19%.
- A 10% platform fee, applied to the amount including VAT.
VAT = credits × 19%platform fee = (credits + VAT) × 10%total paid = credits × 1.19 × 1.10 = credits × 1.309| Credits | VAT 19% | Fee 10% | Total paid |
|---|---|---|---|
| 20.000 TND | 3.800 TND | 2.380 TND | 26.180 TND |
| 50.000 TND | 9.500 TND | 5.950 TND | 65.450 TND |
| 100.000 TND | 19.000 TND | 11.900 TND | 130.900 TND |
| 200.000 TND | 38.000 TND | 23.800 TND | 261.800 TND |
| 500.000 TND | 95.000 TND | 59.500 TND | 654.500 TND |
VAT and the platform fee are charged once, at top-up. Model prices and the amounts deducted per request do not include any further tax or fee.
Invoices
Every completed top-up has an invoice in TND, available from the transaction history on the Billing page, ready to print or save as PDF. Businesses that need a specific billing name, tax id or a payment link for their accounting department can contact us.
Controlling spend
- Per-key monthly limits. Give each key a limit in TND. Once a key has spent it in the current month (UTC), its requests return
402withkey_monthly_limituntil the next month. See Authentication. - max_tokens. Cap the length, and therefore the cost, of each answer.
- Usage dashboard. Usage shows spend per day, per model and per key, with every request and a CSV export.
Looking up the cost of a request
Pass the id of a chat completion (the id field of the response, or of any stream chunk) to get its token usage and the exact amount deducted, with the same API key authentication.
curl "https://promptix.tn/api/v1/generation?id=gen-1758880023-kQ3vX8mZpL2nR7tW" \ -H "Authorization: Bearer $PROMPTIX_API_KEY"{ "data": { "id": "gen-1758880023-kQ3vX8mZpL2nR7tW", "model": "openai/gpt-4o-mini", "created_at": "2026-09-26T09:47:03.512Z", "usage": { "prompt_tokens": 31, "completion_tokens": 22, "total_tokens": 53 }, "native_cost": { "prompt": 0.0000404, "completion": 0.0000287, "total": 0.0000691 }, "total_cost": 0.00001785, "metadata": { "referrer": null, "site_title": null } }}| Field | Description |
|---|---|
id | The generation id you looked up. |
model | Model used. |
created_at | When the request was billed (ISO 8601, UTC). |
usage | Prompt, completion and total tokens billed. |
native_cost.total | Amount deducted from your balance, in TND. |
native_cost.prompt / .completion | The total split in proportion to prompt and completion tokens. |
total_cost | Cost reported by the upstream provider, in US dollars. For reference only; you are charged native_cost.total. |
Billing runs right after the response is delivered, so a lookup made immediately can return 404 with {"error": "Generation not found"}. Wait a couple of seconds and retry. Only requests made by your own account can be looked up. A missing id parameter returns 400.
Alternative: GET /usage/{id}
The same information is also available at /usage/{id}, with a billing.status field and the amount in billing.credits_used (TND).
curl https://promptix.tn/api/v1/usage/gen-1758880023-kQ3vX8mZpL2nR7tW \ -H "Authorization: Bearer $PROMPTIX_API_KEY"{ "requestId": "gen-1758880023-kQ3vX8mZpL2nR7tW", "model": "openai/gpt-4o-mini", "usage": { "prompt_tokens": 31, "completion_tokens": 22, "total_tokens": 53 }, "billing": { "status": "completed", "credits_used": 0.0000691, "openrouter_cost_usd": 0.00001785, "message": "Billing completed successfully." }, "created_at": "2026-09-26T09:47:03.512Z"}Common questions
Do I need an international card?
No. Payment goes through Flouci, in dinars, with a Tunisian payment method.
Do free models cost anything?
No. Requests to free models are not charged, but you still need a balance of at least 0.100 TND to send requests.
Am I billed for failed requests?
Requests rejected before they reach the model (invalid body, bad key, insufficient balance, key limit) are never billed. A request that produced tokens is billed for those tokens, including a stream that was interrupted.