Toman pricing, billed by what you use
No subscription, no setup fee, no monthly minimum. Each model carries its final toman price on its own page, and every request is billed against the tokens it actually consumed.
50,000 toman
Up to 50,000,000 toman in a single transaction
0 toman
No subscription, no onboarding fee
0 toman
The temporary hold is released immediately
50 requests a day
Shared capacity, good for testing
A few models people ask about
Toman per 1M tokens. The last column estimates one 1,000-word round trip.
| Model | Input / 1M tokens | Output / 1M tokens | ≈ 1,000 words |
|---|---|---|---|
| GPT-5openai/gpt-5 | 362,656 | 2,901,250 | 4,243 |
| GPT-5 Miniopenai/gpt-5-mini | 72,531 | 580,250 | 849 |
| Claude Sonnet 4.5anthropic/claude-sonnet-4.5 | 870,375 | 4,351,875 | 6,789 |
 Gemini 2.5 Flashgoogle/gemini-2.5-flash | 87,038 | 725,313 | 1,056 |
 DeepSeek V3deepseek/deepseek-chat | 92,840 | 258,211 | 456 |
 Qwen3 Coder 480B A35Bqwen/qwen3-coder | 87,038 | 290,125 | 490 |
| Grok 4.20x-ai/grok-4.20 | 362,656 | 725,313 | 1,414 |
 Mistral Medium 3mistralai/mistral-medium-3 | 116,050 | 580,250 | 905 |
Cheapest models for high-volume work
If you send thousands of requests a day, these give you the most output per toman.
What happens on each request
- 1. Before the call, a temporary hold is placed on your wallet so we know the balance covers it.
- 2. The request goes to the model and the answer streams back to you.
- 3. When it finishes, the real cost is charged and the rest of the hold is released.
- 4. If the request errors or the connection drops, only real usage is charged — usually nothing.
- 5. The exact figure for every request appears in the usage panel and in the response header (
X-Uttapen-Cost-Toman).
Top-ups, invoices and spend control
- Top up with an Iranian bank card; the full amount lands in your wallet.
- Every successful top-up gets an invoice with a unique number, using the legal details from your account settings.
- Each API key can carry a monthly ceiling, an allow-list of models and its own rate limit.
- Usage broken down by day, model and key, in the panel and through
GET /v1/uttapen/usage.
Frequently asked
- How is the cost calculated?
- Each model page carries its final toman price per 1M input and output tokens. A request is billed against the tokens it actually used — no rounding up to a bundle, no minimum per call.
- Am I charged for a failed request?
- No. A temporary hold goes on the wallet before the call and only the real cost is charged after a successful answer. If it errors or the connection drops, the hold is released.
- Is there a subscription or a fixed monthly fee?
- No. You pay for usage only, and wallet credit stays where it is until you spend it.
- What is the minimum top-up and how do I pay?
- 50,000 toman, with any Iranian bank card through the payment gateway. Every top-up produces an invoice with a unique number.
- Are prices fixed?
- A model's price changes when market conditions or the provider's own price change. The live figure is always on the model page and in the catalogue, with the date it was last updated.
- Are the free models really free?
- Yes, but their capacity is shared and each account has a daily cap. For anything people depend on, a paid model is steadier and faster.
- Can I cap what a project spends?
- Yes. Each API key can have a monthly toman ceiling, an allow-list of models and its own rate limit, and each key's spend is reported separately.
base_url = https://api.uttapen.ir/v1