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

# Inferoute Frequently Asked Questions

> Answers to common questions about Inferoute: SDK compatibility, supported providers, billing, data handling, latency overhead, streaming, and how routing works.

This page answers the most common questions about Inferoute. If you cannot find what you are looking for here, reach out via the dashboard chat or email [support@inferoute.ai](mailto:support@inferoute.ai).

<AccordionGroup>
  <Accordion title="Is Inferoute compatible with the OpenAI SDK?">
    Yes. Inferoute exposes an OpenAI-compatible API, so you can use the official OpenAI SDK by pointing it at the Inferoute base URL and supplying your Inferoute API key:

    ```python python theme={null}
    import openai

    client = openai.OpenAI(
        base_url="https://api.inferoute.ai/v1",
        api_key="YOUR_INFEROUTE_API_KEY",
    )
    ```

    ```javascript node.js theme={null}
    import OpenAI from "openai";

    const client = new OpenAI({
      baseURL: "https://api.inferoute.ai/v1",
      apiKey: process.env.INFEROUTE_API_KEY,
    });
    ```

    No other code changes are required.
  </Accordion>

  <Accordion title="Which providers does Inferoute support?">
    Inferoute currently routes requests across the following providers:

    * **OpenAI** — GPT-4o, GPT-4o mini, o1, o3, and more
    * **Anthropic** — Claude 3.5 Sonnet, Claude 3 Opus, Claude 3 Haiku, and more
    * **Google** — Gemini 1.5 Pro, Gemini 1.5 Flash, and more
    * **Mistral** — Mistral Large, Mistral Small, Codestral, and more
    * **Meta / Llama** — Llama 3.3, Llama 3.1, and more

    Call `GET /v1/models` to retrieve the current list of available models.
  </Accordion>

  <Accordion title="Do I need accounts with each provider?">
    No. Inferoute handles provider relationships on your behalf. You only need a Inferoute account and a Inferoute API key. You do not need to create accounts with OpenAI, Anthropic, Google, or any other provider, and you do not need to manage separate provider API keys.
  </Accordion>

  <Accordion title="How is billing calculated?">
    Your bill is based on two components:

    1. **Provider token costs** — the cost of the tokens consumed at the provider level, passed through at the provider's published rates.
    2. **Inferoute routing fee** — a small per-request fee that covers routing infrastructure, fallback logic, and platform features.

    Both components are aggregated and billed monthly. You can view a breakdown by key, model, and provider in the **Usage** section of the dashboard.
  </Accordion>

  <Accordion title="What happens if a provider goes down?">
    Inferoute monitors provider availability continuously. If a provider becomes unavailable or returns errors above a threshold, Inferoute automatically routes your request to a backup provider that serves an equivalent model. This fallback happens transparently — your application receives a successful response without any code changes. You can configure fallback behavior and preferred backup providers in the **Routing** section of the dashboard.
  </Accordion>

  <Accordion title="Can I use a specific provider exclusively?">
    Yes. Use provider-prefixed model names in your request to pin traffic to a specific provider. For example:

    ```python python theme={null}
    response = client.chat.completions.create(
        model="openai/gpt-4o",  # always use OpenAI, never fall back
        messages=[{"role": "user", "content": "Hello!"}],
    )
    ```

    Supported prefixes include `openai/`, `anthropic/`, `google/`, `mistral/`, and `meta/`. When you use a prefixed model name, Inferoute does not attempt automatic fallback to another provider.
  </Accordion>

  <Accordion title="Is my data stored by Inferoute?">
    By default, Inferoute does not store the content of your requests or responses. Requests are routed to the selected provider and the response is forwarded back to you. Inferoute retains metadata (such as token counts, latency, and provider used) for billing and analytics purposes. If you need data residency guarantees or a custom retention policy, contact [support@inferoute.ai](mailto:support@inferoute.ai) to discuss Enterprise options.
  </Accordion>

  <Accordion title="What is the latency overhead of routing?">
    The routing layer typically adds less than 10 milliseconds to end-to-end latency. The dominant factor in response time is the provider's own inference time, not the Inferoute routing step. If you need to minimize latency further, enable the **latency-optimized** routing strategy in the dashboard, which selects the provider with the lowest observed response time for each request.
  </Accordion>

  <Accordion title="Can I use streaming responses?">
    Yes. Set `stream: true` in your request exactly as you would with the OpenAI API. Inferoute transparently streams the provider's response back to your client.

    ```python python theme={null}
    stream = client.chat.completions.create(
        model="gpt-4o",
        messages=[{"role": "user", "content": "Tell me a story."}],
        stream=True,
    )

    for chunk in stream:
        print(chunk.choices[0].delta.content or "", end="", flush=True)
    ```
  </Accordion>

  <Accordion title="How do I get support?">
    You can reach the Inferoute support team through two channels:

    * **Dashboard chat** — click the chat icon in the bottom-right corner of any dashboard page for real-time help.
    * **Email** — send a message to [support@inferoute.ai](mailto:support@inferoute.ai). Include the `X-Inferoute-Request-Id` header value from any failing request to help the team diagnose the issue faster.
  </Accordion>
</AccordionGroup>