> ## Documentation Index
> Fetch the complete documentation index at: https://docs.tokengift.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Models & pricing

> Per-token prices by model. Our service fee is included here — there are no other fees, ever.

TokenGift makes money one way: a small service margin included in the per-token prices
below. Gifts and top-ups are **never** skimmed — a $10 gift is $10.00 of balance.

## Supported models

Prices are per **1 million tokens**, in USD, service fee included.

| Model                        | Input   | Output  |
| ---------------------------- | ------- | ------- |
| `openai/gpt-4o-mini`         | \$0.162 | \$0.648 |
| `openai/gpt-4o`              | \$2.70  | \$10.80 |
| `anthropic/claude-haiku-4-5` | \$1.08  | \$5.40  |

<Note>
  The model list is curated and growing. Prices are versioned — the price at request time
  is the price you pay, and changes are announced in the changelog before they take effect.
</Note>

## How billing works per request

1. Before the request runs, we reserve an upper-bound estimate (based on your input size
   and `max_tokens`).
2. The request runs; we meter actual usage from the provider.
3. We settle for the actual cost and instantly return the difference to your wallet.

You can see the exact cost of every request in the `TokenGift-Request-Cost` response
header and in your dashboard's usage table.