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

# Models and thinking effort

> Pick the model the agent uses, set how hard it thinks, and turn web search on or off.

The model menu in the generation input picks which AI model the agent runs on. It also holds the thinking effort setting, which trades speed for depth.

Seven of the nine models work on every plan. The two Fable models need a paid plan. Every model uses credits.

![The model menu open in the generation input](https://mintcdn.com/polymet/TfIzLh80_L4bW9xQ/images/docs/canvas/model-menu.png?fit=max\&auto=format\&n=TfIzLh80_L4bW9xQ\&q=85\&s=8b7d3dadfcebf3110e105615b132bc9e)

## Pick a model

1. Click the model name in the generation input.
2. Pick a model from the list. Your next prompt uses it.
3. Polymet asks whether to make it your default, for example **Use Claude Sonnet 5.5 as your default model?** Click **Set as default** to start every new thread with it, or **Not now** to use it in this thread only.

You can also type `/model` in the generation input, then part of a model's name, and pick from the list.

The model can't be changed while the agent is waiting for you to approve an action. The menu says **Model is locked during tool confirmation** until you answer.

## The models

| Model | Plans |
| - | - |
| Claude Sonnet 5 | Every plan |
| Claude Sonnet 5.5 | Every plan |
| Claude Opus 5 | Every plan |
| Claude Opus 5.5 | Every plan. The default for new users |
| Claude Fable 5 | Paid plans |
| Claude Fable 5.1 | Paid plans |
| Gemini 4 Argon | Every plan |
| Gemini 3.1 Pro | Every plan |
| Gemini 3.8 Flash | Every plan |

On the Free plan, the Fable models are greyed out and marked **Paid plans**. Every paid plan includes them, Keep Alive too. See [Plans](https://docs.polymet.ai/plans/plans.md).

If your default model is one your plan no longer includes, the model button shows a lock. Pick another model before you send.

Credits are charged by AI model usage, so a model with higher rates, a higher thinking effort or a longer generation spends more. See [How credits work](https://docs.polymet.ai/plans/credits.md).

## Thinking effort

Open the model menu and drag the **Thinking effort** slider between **Faster** and **Smarter**:

| Level | What it is for |
| - | - |
| **Low** | Fastest and cheapest. Fine for edits and small changes |
| **Medium** | Balanced speed and depth. Right for most work. This is where it starts |
| **High** | Slower and costs more per message. Better on hard problems |

The setting is yours, not your project's. It applies to all your generations, on every model and in every project.

## Web search

The globe button in the generation input lets the agent search and read the web while it works, for example to look up a library's docs or read a page you name. It is on to start with.

* On: **Web search is on — the model can search and read the web**
* Off: **Web search is off — the model can't search the web**

Like thinking effort, this setting is yours and applies in every project.

## Related

* [Prompt the agent](https://docs.polymet.ai/canvas/prompt.md) — the rest of the generation input
* [Threads, context and queued messages](https://docs.polymet.ai/canvas/threads-and-context.md) — how a new thread picks up your default model
* [How credits work](https://docs.polymet.ai/plans/credits.md) — what a generation costs
* [Plans](https://docs.polymet.ai/plans/plans.md) — which plan includes the Fable models


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.