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

# How Billing Works

> Understand Routeway subscriptions, pay-as-you-go pricing, and payment methods.

Routeway offers two billing models: **Pay-As-You-Go** for flexible, usage-based access to every model, and **DeepSeek Subscriptions** for teams with consistent DeepSeek workloads.

<CardGroup cols={2}>
  <Card title="Pay-As-You-Go" icon="bolt" href="#pay-as-you-go">
    Access every model. Pay only for what you use — no commitments, no minimums.
  </Card>

  <Card title="DeepSeek Subscriptions" icon="calendar" href="#deepseek-subscriptions">
    Reserved token budgets and bonus perks for consistent DeepSeek usage.
  </Card>
</CardGroup>

***

## Pay-As-You-Go

PAYG gives you access to every model on the Routeway API — OpenAI, Anthropic, Gemini, DeepSeek, Llama, and more. You're charged only for the tokens you process, with no monthly commitment required.

<AccordionGroup>
  <Accordion title="How tokens are billed" icon="coins" defaultOpen>
    Every request is billed in two parts:

    | Component         | What it covers                      |
    | :---------------- | :---------------------------------- |
    | **Input tokens**  | The prompt and any context you send |
    | **Output tokens** | The text generated in the response  |

    Rates are quoted **per 1 million tokens** and vary by model. You can view live rates in the [Models Catalog](https://routeway.ai/models) or query them via the API.
  </Accordion>

  <Accordion title="Service tiers" icon="gauge">
    Some models support **flex** and **priority** processing tiers in addition to the default tier. These are provider-side features from OpenAI and Gemini — Routeway passes your tier through to the upstream provider and bills you at the corresponding rate. Routeway does not run its own queue or priority routing.

    | Tier         | Cost             | Trade-off                                            |
    | :----------- | :--------------- | :--------------------------------------------------- |
    | **Default**  | Standard pricing | Balanced TTFT                                        |
    | **Flex**     | 50% off          | Same output quality; TTFT may occasionally be higher |
    | **Priority** | 2× standard      | Lower, more consistent TTFT                          |

    Set a tier via the `service_tier` request parameter, or append `:flex` / `:priority` to the model ID. Currently available on select OpenAI and Gemini models only — check `capabilities.service_tiers` in the model object.

    See [Service Tiers](/guides/chat-completions/service-tiers) for full details.
  </Accordion>

  <Accordion title="Prompt caching" icon="database">
    Caching support varies by model. Check the `pricing.caching` field returned by the API (or the [Models Catalog](https://routeway.ai/models)) to see what a specific model supports.

    | Tier                        | When you're charged                                    | Billing key        |
    | :-------------------------- | :----------------------------------------------------- | :----------------- |
    | **Cache read**              | A cached prompt is reused (cheaper than a fresh input) | `caching.read`     |
    | **Cache write (5 min TTL)** | A prompt is written into cache, kept for 5 minutes     | `caching.write.5m` |
    | **Cache write (1 hr TTL)**  | A prompt is written into cache, kept for 1 hour        | `caching.write.1h` |

    Not every model has every tier:

    * **Read-only** (e.g. Kimi K2.7 Code) — cache hits are discounted; there is no explicit write charge.
    * **Read + write** (e.g. Anthropic models) — both reads and writes are billed, with write rates varying by TTL.
    * **No caching** — `pricing.caching` is absent from the model's pricing object.
  </Accordion>
</AccordionGroup>

***

## DeepSeek Subscriptions

Subscriptions include a reserved DeepSeek token budget and free daily model requests across all plans.

<Tabs>
  <Tab title="Core — $6.99 / $15.99">
    |                              | Biweekly  | Monthly   |
    | :--------------------------- | :-------- | :-------- |
    | **Price**                    | \$6.99    | \$15.99   |
    | **Included DeepSeek tokens** | 15M       | 35M       |
    | **Free model requests**      | 250 / day | 250 / day |
  </Tab>

  <Tab title="Core Pro — $12.99 / $34.99">
    |                              | Biweekly  | Monthly   |
    | :--------------------------- | :-------- | :-------- |
    | **Price**                    | \$12.99   | \$34.99   |
    | **Included DeepSeek tokens** | 28M       | 80M       |
    | **Free model requests**      | 300 / day | 330 / day |
    | **PAYG text discount**       | 5% off    | 5% off    |

    <Note>Core Pro includes an additional **5% discount** on all Pay-As-You-Go text usage.</Note>
  </Tab>
</Tabs>

<Info>
  Subscription tokens are consumed first. Once exhausted, usage automatically continues on PAYG (unless you disable key fallback in the dashboard).
</Info>

***

## Cost Examples

Routeway calculates your cost using the standard token billing formula:

$\text{Total Cost} = (\text{Input Tokens} \times \text{Input Rate}) + (\text{Output Tokens} \times \text{Output Rate})$

<Tabs>
  <Tab title="No caching">
    **`deepseek-v3`** — $0.23/1M input · $0.64/1M output

    |                        | Tokens    | Cost          |
    | :--------------------- | :-------- | :------------ |
    | Input (≈ 750 words)    | 1,000     | \$0.00023     |
    | Output (≈ 1,500 words) | 2,000     | \$0.00128     |
    | **Total**              | **3,000** | **\$0.00151** |
  </Tab>

  <Tab title="Cache read only">
    **`kimi-k2.7-code`** — $0.95/1M input · $4.00/1M output · \$0.20/1M cache read

    Suppose 800 of your 1,000 input tokens were already cached:

    |             | Tokens | Rate      | Cost          |
    | :---------- | :----- | :-------- | :------------ |
    | Fresh input | 200    | \$0.95/1M | \$0.00019     |
    | Cache read  | 800    | \$0.20/1M | \$0.00016     |
    | Output      | 2,000  | \$4.00/1M | \$0.00800     |
    | **Total**   |        |           | **\$0.00835** |

    Without caching the same request would cost **\$0.00950** — a \~12% saving.
  </Tab>

  <Tab title="Cache read + write">
    **`claude-fable-5`** — $10.00/1M input · $50.00/1M output · $1.00/1M cache read · $12.50/1M cache write (5 min)

    First request — prompt is written into cache:

    |                     | Tokens | Rate       | Cost          |
    | :------------------ | :----- | :--------- | :------------ |
    | Cache write (5 min) | 1,000  | \$12.50/1M | \$0.01250     |
    | Output              | 500    | \$50.00/1M | \$0.02500     |
    | **Total**           |        |            | **\$0.03750** |

    Subsequent requests within 5 minutes — same prompt served from cache:

    |            | Tokens | Rate       | Cost          |
    | :--------- | :----- | :--------- | :------------ |
    | Cache read | 1,000  | \$1.00/1M  | \$0.00100     |
    | Output     | 500    | \$50.00/1M | \$0.02500     |
    | **Total**  |        |            | **\$0.02600** |

    Each cache hit saves **\$0.01150** compared to a fresh input read.
  </Tab>
</Tabs>

***

## Live Pricing

View rates anytime in the [Interactive Models Catalog](https://routeway.ai/models), or fetch them programmatically:

```bash theme={null}
curl https://api.routeway.ai/v1/models
```

Each model returns a `pricing` object. The `caching` field is only present when the model supports it, and its shape reflects exactly what that model offers:

<Tabs>
  <Tab title="No caching">
    ```json theme={null}
    {
      "id": "deepseek-v3",
      "pricing": {
        "input": { "unit": "1M tokens", "price_per_million_t": 0.23 },
        "output": { "unit": "1M tokens", "price_per_million_t": 0.64 }
      }
    }
    ```
  </Tab>

  <Tab title="Cache read only">
    ```json theme={null}
    {
      "id": "kimi-k2.7-code",
      "pricing": {
        "input":  { "unit": "1M tokens", "price_per_million_t": 0.95 },
        "output": { "unit": "1M tokens", "price_per_million_t": 4.0 },
        "caching": {
          "read": { "unit": "1M tokens", "price_per_million_t": 0.2 }
        }
      }
    }
    ```
  </Tab>

  <Tab title="Cache read + write">
    ```json theme={null}
    {
      "id": "claude-fable-5",
      "pricing": {
        "input":  { "unit": "1M tokens", "price_per_million_t": 10.0 },
        "output": { "unit": "1M tokens", "price_per_million_t": 50.0 },
        "caching": {
          "read": { "unit": "1M tokens", "price_per_million_t": 1.0 },
          "write": {
            "5m": { "unit": "1M tokens", "price_per_million_t": 12.5 },
            "1h": { "unit": "1M tokens", "price_per_million_t": 20.0 }
          }
        }
      }
    }
    ```
  </Tab>
</Tabs>

***

## Payment Methods

Payments are processed securely through **[Polar.sh](https://polar.sh)**:

* Credit / Debit cards (Visa, Mastercard, Amex, and more)
* Automatic invoicing

<Note>
  **Your card details never touch Routeway's servers.** All transactions are handled exclusively by Polar.sh, which is fully PCI-compliant.
</Note>

<Accordion title="Do you support PayPal or other payment methods?">
  **PayPal is supported**, though it isn't available by default. For other payment methods, availability may vary — reach out to us and we'll let you know what's possible:

  * **Discord**: Join [discord.gg/RjX2CpdPpd](https://discord.gg/RjX2CpdPpd) and open a ticket — this is the fastest way to reach us.
  * **Email**: [help@team.routeway.ai](mailto:help@team.routeway.ai)
</Accordion>