# Effort



<DocImage src="features/effort/effort-selector" alt="The Effort selector next to the model picker, open with Default, None, Low, Medium, High and Xhigh" width="840" height="520" />

Effort controls how much reasoning a supported model uses before it answers. Higher effort means more thorough responses, but takes longer and uses your limits faster.

Use the Effort selector next to the model picker when you want to trade speed for deeper reasoning.

## Effort Levels [#effort-levels]

| Effort     | Best for                                           |
| ---------- | -------------------------------------------------- |
| **Low**    | Fastest responses with minimal reasoning.          |
| **Medium** | Balanced reasoning for everyday tasks.             |
| **High**   | Deeper reasoning for more complex work.            |
| **Xhigh**  | Extended reasoning for the hardest problems.       |
| **Max**    | Maximum reasoning depth. Highest cost and slowest. |

## When to Use It [#when-to-use-it]

* Use **Low** or **Medium** for quick edits, straightforward questions, and normal app-building work.
* Use **High** when the task needs deeper planning, debugging, architecture, or multi-step reasoning.
* Use **Xhigh** or **Max** only for the hardest problems where extra time and usage are worth it.

<Callout type="info">
  Not every provider exposes the same Effort options. Ideavo only shows options
  supported by the selected model.
</Callout>

## Provider Support [#provider-support]

Effort availability depends on both the provider and the model:

* **OpenAI / GPT-style models** commonly support `low`, `medium`, and `high`. Some newer GPT or Codex models may also support `minimal`, `none`, or `xhigh`.
* **Anthropic models** expose `high` and `max` when reasoning is supported.
* **Google / Gemini models** vary by model, commonly exposing `low` / `high` or `high` / `max`.
* Other providers may expose fewer options or no Effort selector at all.

## Custom Models [#custom-models]

For custom or local models, Effort appears only when the model configuration has reasoning enabled.

<Callout type="warn">
  Only enable reasoning for a custom model if the model and provider API truly
  support reasoning or effort controls. Enabling it for unsupported models can
  cause failed requests or unexpected responses.
</Callout>

See [Custom Provider](/docs/features/custom-provider) for custom provider setup.

## FAQ [#faq]

<Accordions>
  <Accordion title="Why don't I see the Effort selector?">
    Not all models support Effort. Ideavo only shows the selector when the
    selected model and provider expose supported reasoning effort options.
  </Accordion>

  <Accordion title="Does higher effort always mean better output?">
    Not always. Higher effort is useful for complex reasoning, but simple tasks
    are usually faster and cheaper with Low, Medium, or Default.
  </Accordion>

  <Accordion title="Can I change Effort between messages?">
    Yes. You can change Effort before sending your next prompt. The new setting
    applies to future responses from the selected model.
  </Accordion>

  <Accordion title="Does Effort affect speed and usage?">
    Yes. Higher Effort can take longer and use your limits faster because the
    model spends more work on reasoning before producing the final response.
  </Accordion>

  <Accordion title="Should I enable reasoning for every custom model?">
    No. Enable reasoning only for custom models that explicitly support
    reasoning or effort controls. Unsupported models may fail or behave
    unpredictably.
  </Accordion>
</Accordions>
