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

# Compare Providers by Cost

> Send the same prompt to different providers through one URL and see which one is cheapest.

**Run the same workload across multiple providers and let the dashboard show you the cost difference.**

<Steps>
  <Step title="Send requests to different providers through the same base URL">
    Toolken's gateway detects the provider from the model name. To compare, send the same prompt using a model from each provider you want to evaluate. You do not need separate SDKs or base URLs.

    For OpenAI-compatible providers (OpenAI, Groq, Mistral, Deepseek, and others), use `/v1/chat/completions`. For Anthropic models, use `/v1/messages`.

    <CodeGroup>
      ```python Python theme={null}
      from openai import OpenAI

      client = OpenAI(
          base_url="https://gateway.toolken.ai/v1",
          api_key="sk-...",  # your provider key, forwarded untouched
          default_headers={
              "X-Toolken-Key": "tk_live_...",
              "X-Toolken-Metadata-Feature": "provider-comparison",
          },
      )

      prompt = [{"role": "user", "content": "Explain backpressure in stream processing in two sentences."}]

      # OpenAI
      client.chat.completions.create(model="gpt-4o-mini", messages=prompt)

      # Groq (OpenAI-compatible, same endpoint)
      client_groq = OpenAI(
          base_url="https://gateway.toolken.ai/v1",
          api_key="gsk_...",  # your Groq key
          default_headers={
              "X-Toolken-Key": "tk_live_...",
              "X-Toolken-Metadata-Feature": "provider-comparison",
          },
      )
      client_groq.chat.completions.create(model="llama-3.1-8b-instant", messages=prompt)
      ```

      ```javascript Node theme={null}
      import OpenAI from "openai";

      const prompt = [{ role: "user", content: "Explain backpressure in stream processing in two sentences." }];

      // OpenAI
      const openaiClient = new OpenAI({
        baseURL: "https://gateway.toolken.ai/v1",
        apiKey: process.env.OPENAI_API_KEY, // forwarded untouched
        defaultHeaders: {
          "X-Toolken-Key": "tk_live_...",
          "X-Toolken-Metadata-Feature": "provider-comparison",
        },
      });
      await openaiClient.chat.completions.create({ model: "gpt-4o-mini", messages: prompt });

      // Groq (OpenAI-compatible, same endpoint)
      const groqClient = new OpenAI({
        baseURL: "https://gateway.toolken.ai/v1",
        apiKey: process.env.GROQ_API_KEY,
        defaultHeaders: {
          "X-Toolken-Key": "tk_live_...",
          "X-Toolken-Metadata-Feature": "provider-comparison",
        },
      });
      await groqClient.chat.completions.create({ model: "llama-3.1-8b-instant", messages: groqMessages });
      ```

      ```bash curl theme={null}
      PROMPT='{"model":"gpt-4o-mini","messages":[{"role":"user","content":"Explain backpressure in stream processing in two sentences."}]}'

      # OpenAI (gpt-4o-mini)
      curl https://gateway.toolken.ai/v1/chat/completions \
        -H "Authorization: Bearer $OPENAI_API_KEY" \
        -H "X-Toolken-Key: tk_live_..." \
        -H "X-Toolken-Metadata-Feature: provider-comparison" \
        -H "Content-Type: application/json" \
        -d "$PROMPT"

      # Groq (OpenAI-compatible, same endpoint, different model)
      curl https://gateway.toolken.ai/v1/chat/completions \
        -H "Authorization: Bearer $GROQ_API_KEY" \
        -H "X-Toolken-Key: tk_live_..." \
        -H "X-Toolken-Metadata-Feature: provider-comparison" \
        -H "Content-Type: application/json" \
        -d '{"model":"llama-3.1-8b-instant","messages":[{"role":"user","content":"Explain backpressure in stream processing in two sentences."}]}'
      ```
    </CodeGroup>

    <Note>Anthropic models use a different endpoint (`/v1/messages`) and a different auth header (`x-api-key`). You cannot compare an Anthropic model and an OpenAI model in the same SDK call, but Toolken records both and you can compare costs in the dashboard.</Note>
  </Step>

  <Step title="Open the dashboard and group by Provider or Model">
    Go to the cost view. Group by **Model** to see per-model cost side by side, or group by **Provider** for a higher-level view. Filter by the `provider-comparison` feature to isolate the requests from this experiment.

    <Frame caption="Cost by model in the overview">
      <img src="https://mintcdn.com/toolken/B9FQGhQVN-q79Dqo/images/dashboard-overview.png?fit=max&auto=format&n=B9FQGhQVN-q79Dqo&q=85&s=6cea8a13b610ff1cddcc6ecc04c04cef" alt="Dashboard showing cost broken down by model" width="1440" height="900" data-path="images/dashboard-overview.png" />
    </Frame>
  </Step>
</Steps>

<AccordionGroup>
  <Accordion title="Reach more models with OpenRouter">
    OpenRouter aggregates a wide catalog of models from many providers. Set `X-Toolken-Provider: openrouter` and pass your OpenRouter key. You can then use any model in their catalog without signing up for each provider separately.

    ```python theme={null}
    default_headers={
        "X-Toolken-Key": "tk_live_...",
        "X-Toolken-Provider": "openrouter",
        "X-Toolken-Metadata-Feature": "provider-comparison",
    }
    ```

    The dashboard attributes each request to its actual model, so cost comparisons still work.
  </Accordion>

  <Accordion title="Force a specific provider with X-Toolken-Provider">
    By default the gateway detects the provider from the model name. To override that detection, set `X-Toolken-Provider` explicitly. This is useful when two providers accept the same model name, or when you want to test the same model on different infrastructure.

    ```bash theme={null}
    curl https://gateway.toolken.ai/v1/chat/completions \
      -H "Authorization: Bearer $YOUR_KEY" \
      -H "X-Toolken-Key: tk_live_..." \
      -H "X-Toolken-Provider: groq" \
      -H "Content-Type: application/json" \
      -d '{"model":"llama-3.1-8b-instant","messages":[{"role":"user","content":"Hello"}]}'
    ```
  </Accordion>
</AccordionGroup>

## Next

<CardGroup cols={2}>
  <Card title="Cost Per Agent" icon="robot" href="/use-cases/cost-per-agent">Break down spend by agent or workflow.</Card>
  <Card title="Cost by Environment" icon="server" href="/use-cases/cost-by-environment">Separate production and staging spend.</Card>
</CardGroup>