Pricing and benchmark ยท 2026-05-21
AI API cost in Turkey: token prices and credit margin
How to read AI API cost across token pricing, input-output split, free models, and LLMTR's 8% platform margin on credit top-ups.
Where cost calculation starts
LLM API cost has two separate surfaces. The first is the model's input and output token price. The second is any transaction or platform margin applied when adding credits.
On LLMTR, model prices are preserved as catalog values. The 8% platform margin is applied to the credit top-up amount, not to model token prices.
Example credit calculation
For a 100 USD credit top-up, the credited amount is 100 USD and the total charged amount is 108 USD with the 8% platform margin. This distinction prevents model-page token prices from being misread.
- Credit: 100 USD.
- Platform margin: 8 USD.
- Charge: 108 USD.
- Model usage price: calculated separately from the selected model catalog row.
What to measure in a benchmark
Looking only at input token price gives an incomplete comparison. Output cost, context size, reasoning behavior, and real tokens per request should be measured together.
Frequently asked questions
Is the 8% margin added to model prices?
No. The margin is added to the credit top-up amount. Model prices remain separate catalog values.
Are free models enough for production?
Free models can be useful for trials and low-risk workloads. For production, quality, continuity, data policy, and endpoint support still need review.