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

# Anthropic

> Use your Anthropic API key to run Claude models inside Rumus.

Configure your own Anthropic key to use Claude Opus, Sonnet, and Haiku directly. Requests go from your machine to Anthropic — Rumus does not proxy them.

## Before you start

You need:

* An Anthropic account and an API key from the [Anthropic Console](https://console.anthropic.com/).
* Available credit on your Anthropic account.

## Add Anthropic in Rumus

<Steps>
  <Step title="Open the model settings">
    Go to **Settings → AI → Models** and click **Add Model**.
  </Step>

  <Step title="Pick the provider">
    Set **Provider** to **Anthropic**.
  </Step>

  <Step title="Paste your API key">
    Paste the key from the Anthropic Console into **API Key**. The field is masked and stored encrypted in your local [vault](/docs/account/vault).
  </Step>

  <Step title="(Optional) Custom base URL">
    Leave **Base URL** blank to use the default (`https://api.anthropic.com`). Override it only if you're routing through an Anthropic-compatible proxy or regional endpoint.
  </Step>

  <Step title="Pick a model">
    Choose from the list (Opus 4.6, Sonnet 4.6, Haiku 4.5, etc.) or toggle **Enter custom ID** to type a model ID manually — useful for snapshot or preview models.
  </Step>

  <Step title="Capabilities">
    Switch to the **Capabilities** tab and enable what the model supports. For Claude models you typically want:

    * **Tool Calling** — required for agentic mode.
    * **Vision** — for image input on multimodal Claude versions.
    * **Prompt Cache** — Claude supports prompt caching; leave on to cut cost on repeated prompts.
  </Step>

  <Step title="Save">
    Click **Save**. The model now appears in the chat model picker under **Custom Models**.
  </Step>
</Steps>

## Recommended models

| Model                 | Good for                                                                 |
| --------------------- | ------------------------------------------------------------------------ |
| **Claude Opus 4.6**   | Hardest tasks: tricky debugging, multi-step refactors, careful reasoning |
| **Claude Sonnet 4.6** | Default daily driver — fast, smart, agentic                              |
| **Claude Haiku 4.5**  | High-volume, latency-sensitive work like autocomplete                    |

For Anthropic's full lineup and capabilities, see the [Anthropic model docs](https://docs.anthropic.com/).

## Tips

* **Prompt caching** is on by default for Claude. With long system prompts or repeated context, cache hits can drop input cost by \~90%.
* **Extended thinking** ("reasoning") is supported. Toggle it per thread from the model picker; Anthropic bills thinking tokens as output tokens.
* If you hit rate limits, raise them in your Anthropic Console workspace settings — Rumus surfaces the upstream error verbatim.

## Troubleshooting

<AccordionGroup>
  <Accordion title="401 Invalid API Key">
    Recheck the key in the Anthropic Console. Anthropic keys start with `sk-ant-`. Make sure you didn't paste a workspace ID by mistake.
  </Accordion>

  <Accordion title="429 Rate limited">
    You've hit Anthropic's per-minute or per-day limits. Wait, switch to a smaller model, or raise limits on your Anthropic workspace.
  </Accordion>

  <Accordion title="Model not in the list">
    Use **Enter custom ID** to type the model ID exactly (e.g. `claude-opus-4-6-20251115`).
  </Accordion>
</AccordionGroup>

<Note>
  Hit a snag we didn't cover? Ask in the [Rumus community](https://www.rumus.ai/community).
</Note>

## Next steps

<CardGroup cols={2}>
  <Card title="Other providers" icon="grid" href="/docs/models/overview">
    OpenAI, Google, Z.AI, DeepSeek, Kimi, Ollama, OpenAI-compatible.
  </Card>

  <Card title="AI assistant" icon="robot" href="/docs/ai/overview">
    Plan mode, sub-agents, MCP, command approval.
  </Card>
</CardGroup>