---
title: "Choose a model and thinking level"
description: "Find the model selected for your session, check the inputs it supports, and adjust its thinking level for a research task."
---

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

# Choose a model and thinking level

The model is the AI system that reads your request and produces a response. The thinking level controls how much reasoning effort the model uses for that request.

Open your study project and a conversation. If you need to connect your account, follow [Sign in to the desktop app](/get-started/sign-in-to-desktop/).

## Select a model

1. Select the model name in the request box. In the example, **GPT Terra** is selected and has a check mark.

   ![Model picker with GPT Terra selected and GPT Luna marked Default](@/assets/images/work-with-agents/agent-model-current.png)

2. Type part of a name in **Search models**. For example, type `GPT Luna` to narrow the list to choices with that name.

   ![Search models filtered to GPT Luna, GPT Luna Fast, and GPT Luna Flex](@/assets/images/work-with-agents/agent-model-search.png)

3. Select a result and confirm that its name appears in the request box.

InsightGig starts with a model and thinking level selected. You can send your first request with those settings.

The **Default** label in the model list identifies the default model. The check mark shows the model this conversation is using. They can appear on different rows.

If you can't find a model, select the sliders icon to check **Manage models**. A managed workspace can also limit which models you see. Follow [Manage model access and budgets](/console/workspace/manage-model-access/) to check both settings.

### Check what a model can read

Point to a model in the picker to see the inputs it supports. For a concept card or a pack design, look for image input before sending the image with your request.

The details can also show a **Context limit**. This is how much material fits into a single model request, measured in tokens. Tokens are pieces of text, and the count includes instructions and conversation history.

For several long transcripts, start with a small set. Ask for quotes and source references, then review the result before expanding the task.

## Choose a thinking level

1. Open the thinking selector beside the model name. It may show the current level, such as **Medium**. Check the available choices for the selected model.

   ![GPT Luna thinking menu showing Default, None, Low, Medium, High, Xhigh, and Max](@/assets/images/work-with-agents/agent-model-thinking.png)

2. Select the level that fits your task. For example, choose **Low** for extracting stated facts from a short interview.

   ![GPT Luna and Low thinking selected in the request box](@/assets/images/work-with-agents/agent-model-low.png)

3. Check the model and thinking level before sending the request. Check them again after you change models or [choose an agent mode](/work-with-agents/agent-controls/choose-agent-mode/).

Start with **Default** if you're unsure. It uses the selected model's default behavior. The menu varies by model, and a higher effort can take longer and use more credits.

## Match the settings to the task

| Research task | Starting point | Check before expanding |
| --- | --- | --- |
| Extract claims from one interview | Try a lower effort on a short transcript with a defined list of fields. | Compare each quote with the transcript and flag missing fields. |
| Compare several interview findings | Try a higher effort when the request involves conflicting evidence or several constraints. | Check that each theme covers exceptions and links to its source. |
| Review packaging or concept images | Choose a model that lists image input. | Compare the text and visual details in the reply with the original image. |

Try the same small task if you need to compare models. Use the same brief and sources, then check which reply best supports the decisions you need to make.

## Next steps

Use [Give agents context](/work-with-agents/chat-with-agents/) to point to your sources before sending. Review the credits used in [Check credits and usage](/console/usage-and-billing/track-usage/).

Source: https://docs.insightgig.ai/work-with-agents/agent-controls/choose-model-and-thinking-level/index.mdx
