# Thinking Modes: Pick the Right Depth for the Task

> Pick how deeply Taskade AI thinks. Leave TSK-Auto on, or choose TSK-Instant, TSK-Standard, TSK-Thinking, or TSK-Reasoning for chat, agents, and app builds.

- Source: https://www.taskade.com/learn/agents/thinking-modes
- Section: AI Agents
- Updated: 2026-10-16
- Index of all pages: https://www.taskade.com/learn/llms.txt

## Overview

Thinking modes set how much thought Taskade AI puts into an answer. In the model menu they appear as **TSK-Auto** and four TSK tiers. TSK-Auto chooses for you. A tier sets a fixed depth, from quick edits to complex apps.

> **TL;DR:** Open the model menu in Taskade EVE, an agent, or the Taskade Genesis prompt box. Leave **TSK-Auto** on, or pick **TSK-Instant**, **TSK-Standard**, **TSK-Thinking**, or **TSK-Reasoning**. TSK-Auto and TSK-Instant work on every plan. TSK-Standard and TSK-Thinking need a paid plan, and TSK-Reasoning needs Business or a higher plan. See [Custom AI Agents](https://www.taskade.com/learn/agents/custom-agents).

## Before You Start

- **What you need**: A Taskade account. Any plan can use TSK-Auto and TSK-Instant.
- **Time**: About one minute.

## Compare the Options

| Option | Menu description | Good for | Plan |
|--------|------------------|----------|------|
| **TSK-Auto** *(default)* | Picks the best model for your task | Everyday work when you do not want to choose | Every plan |
| **TSK-Instant** | Quick edits | Quick edits, short answers, and high-volume chat | Every plan |
| **TSK-Standard** | Everyday apps | Everyday apps, drafts, summaries, and routine answers | Paid plans |
| **TSK-Thinking** | Bigger apps, more care | Bigger apps and longer jobs that need more care | Paid plans |
| **TSK-Reasoning** | Complex apps. Uses more credits | Complex apps, multi-step logic, and hard problems | Business and up |

Deeper tiers use more credits, and TSK-Reasoning uses the most. Hover a tier in the menu to see its credit cost before you send a message. TSK-Auto shows no fixed cost, because its choice changes with the task.

TSK-Thinking came back to the menu in October 2026 with a new model behind it. A TSK-Thinking pick saved before it left the menu in September 2026 still shows **TSK-Auto** and runs on TSK-Auto. Pick **TSK-Thinking** again to use it. A **TSK-Reasoning** pick saved while TSK-Thinking was out of the menu, from 2026-09-28 to 2026-10-03, now shows as **TSK-Thinking**.

## Pick a Tier

1. Open a Taskade EVE chat or the [Taskade Genesis prompt box](https://www.taskade.com/create).
2. Click the model button in the prompt box. Its tooltip reads **AI Model: TSK-Auto** until you pick something else.
3. Hover a tier to see its credit cost.
4. Click **TSK-Auto** or a TSK tier.

The menu looks like this on a Free workspace. A chip marks each tier that needs a higher plan:

```text
┌─ AI Model: TSK-Auto ───────────────────────────────────┐
│ ● TSK-Auto       Recommended                           │
│                  Picks the best model for your task    │
├────────────────────────────────────────────────────────┤
│   TSK-Instant    Quick edits                           │
│   TSK-Standard   Everyday apps              Pro        │
│   TSK-Thinking   Bigger apps, more care     Pro        │
│   TSK-Reasoning  Complex apps. Uses more    Business   │
│                  credits                               │
├────────────────────────────────────────────────────────┤
│   Current Model  ›                                     │
└────────────────────────────────────────────────────────┘
```

Your pick applies to the messages that you send next. You can switch at any time in the same conversation. To pin one exact model instead of a tier, open **Current Model**.

If your plan cannot use a tier, a click on that tier opens the upgrade panel. Your current pick stays in place.

You are done when the model button shows the tier that you picked.

## Match the Tier to the Job

Not sure which option fits the task in front of you? This decision tree points to one.

```mermaid
flowchart TD
    START([What are you working on?]):::start
    START --> Q1{Want to choose<br/>at all?}:::q
    Q1 -->|Let Taskade decide| AUTO[TSK-Auto<br/>recommended default]:::auto
    Q1 -->|I will pick| Q2{How hard is the task?}:::q
    Q2 -->|Quick edit or<br/>short answer| INS[TSK-Instant<br/>fastest, lowest cost]:::ins
    Q2 -->|Everyday app<br/>or draft| STD[TSK-Standard<br/>everyday apps]:::std
    Q2 -->|Bigger app or<br/>longer job| THK[TSK-Thinking<br/>more care]:::thk
    Q2 -->|Complex app or<br/>multi-step logic| RSN[TSK-Reasoning<br/>highest cost]:::rsn
    classDef start fill:#11111b,color:#cdd6f4,stroke:#00d4aa,stroke-width:1.5px
    classDef q fill:#1e1e2e,color:#94a3b8,stroke:#475569,stroke-width:1px
    classDef auto fill:#1a0d24,color:#ce93d8,stroke:#6a1b9a,stroke-width:2px
    classDef ins fill:#1a140a,color:#ffb74d,stroke:#e65100,stroke-width:1.5px
    classDef std fill:#0d1a0e,color:#86efac,stroke:#22c55e,stroke-width:1.5px
    classDef thk fill:#0d1b2a,color:#7dd3fc,stroke:#38bdf8,stroke-width:1.5px
    classDef rsn fill:#2a1225,color:#ff8fa3,stroke:#ff2d60,stroke-width:2px
```

**TSK-Auto** is the right default for most people. It makes the choice for you, so you can focus on the task.

**TSK-Instant** fits a quick rewrite, a short answer, or a routine lookup. It is the fastest tier and uses the fewest credits.

**TSK-Standard** fits everyday apps, drafting, summaries, and routine content.

**TSK-Thinking** fits bigger apps and longer jobs that need more care than TSK-Standard. It uses more credits than TSK-Standard.

**TSK-Reasoning** fits complex apps and step-by-step work, such as a multi-step workflow, a pricing calculation, or a hard research question. It uses the most credits.

## Set a Tier for Agents and Automations

The same options appear wherever you choose a model. The list below the tiers changes with the place.

| Where | How to open the menu | What the menu shows |
|-------|----------------------|---------------------|
| Taskade EVE and the Taskade Genesis prompt box | The model button in the prompt box | TSK-Auto, the TSK tiers, and **Current Model** to pin one model |
| A custom AI agent | The agent's **Knowledge** tab → **AI Model**, or **Edit agent** → **General** | TSK-Auto, the TSK tiers, and **All Models** |
| A Taskade AI step in an automation | **Model (optional)** in the step settings | TSK-Auto and the TSK tiers under a **Taskade credits** heading |

See [Custom AI Agents](https://www.taskade.com/learn/agents/custom-agents) for agent settings and [Taskade AI](https://www.taskade.com/learn/automation/ai-action) for automation steps.

## Frequently Asked Questions

### Do I need to know which model a tier uses?

No. Pick a tier by the job: TSK-Instant for quick edits, TSK-Standard for everyday apps, TSK-Thinking for bigger apps, and TSK-Reasoning for complex apps. If you are not sure, leave TSK-Auto on.

### Do deeper tiers cost more credits?

Yes. Each tier uses more credits than the tier above it in the menu, and TSK-Reasoning uses the most. Hover a tier to see its cost before you send a message. See [AI Models & Credit Costs](https://www.taskade.com/learn/genesis/model-credits).

### Which tiers can a Free workspace use?

A Free workspace can use TSK-Auto and TSK-Instant. TSK-Standard and TSK-Thinking need a paid plan. TSK-Reasoning needs Business or a higher plan. A tier that your plan cannot use shows an upgrade chip.

### Is TSK-Thinking back?

Yes. TSK-Thinking returned to the model menu in October 2026 with a new model behind it, and every paid plan can pick it. A TSK-Thinking pick saved before it left the menu in September 2026 still runs on TSK-Auto, so pick TSK-Thinking again to use it.

### Can an agent keep a fixed tier?

Yes. Open the agent's **Knowledge** tab, open **AI Model**, and pick a tier. The agent then answers with that tier until you change it.

### Is TSK-Auto safe to leave on?

Yes. TSK-Auto is the recommended default. It makes the choice for you on every plan.

## Related guides

- [Custom AI Agents](https://www.taskade.com/learn/agents/custom-agents): Set the tier that an agent uses
- [Taskade EVE](https://www.taskade.com/learn/agents/eve-assistant): The model menu in the Taskade EVE chat
- [AI Models & Credit Costs](https://www.taskade.com/learn/genesis/model-credits): How credits work for each tier
- [Taskade AI](https://www.taskade.com/learn/automation/ai-action): Pick a tier for an automation step
- [TSK-1: The Taskade System Kernel](https://www.taskade.com/learn/genesis/tsk-1): How TSK-Auto fits into Taskade
- [Explore AI agents](https://www.taskade.com/agents)