# Claude Code Models and Effort

> How to pick a Claude Code model by a short name and set how hard it thinks. The keys that switch both without a menu.

**Course:** [Claude Code Basics](https://learn.benibix.com/learn/ai-claude-code)  
**Time:** 9 min

**After this lesson you can:**

- Switch models with /model or an alias
- Raise effort for one prompt with ultrathink
- Know what fast mode trades

## How to pick the model and how hard it thinks

A model alias is a short name for a model. The alias opus picks the newest Claude Opus, and sonnet picks the newest Claude Sonnet. The alias haiku picks Claude Haiku, the small fast one. The alias sonnet[1m] asks for room for one million tokens. Switch with /model during a session, or start with claude --model opus.

The alias opusplan is a mix. It uses Claude Opus while Claude is in plan mode, working out what to do. Then it switches to Claude Sonnet to do the work. You pay for the bigger model only where the hard thinking happens.

The effort level sets how hard the model thinks on each step. Most new models have five levels: low, medium, high, xhigh and max. More effort means slower replies and more tokens. Less is faster and cheaper for simple work. Change it with /effort, or with the arrow keys in the /model menu.

Put the word ultrathink anywhere in a prompt to ask for deeper thinking on that one prompt. Your normal effort comes back on the next one. Claude Code only spots that one word. Phrases like think hard are passed on as plain text, with no special effect.

Fast mode makes a model that supports it answer faster, at a higher price per token. It is the same model, so the quality stays the same. Switch it with /fast, or press Option+O on macOS and Alt+O on Windows and Linux.

Two more keys save trips to a menu. Option+P, or Alt+P, opens the model picker and keeps what you typed in the prompt. Option+T, or Alt+T, turns extended thinking on or off. On the newest models, which always think, that key does nothing.

```console
/model sonnet
/effort high
> ultrathink: plan the move from REST to gRPC
```
Switch to Claude Sonnet, set high effort for the session, then ask for one deeper plan.

![A scale of five effort levels from low to max. The low end is fast and cheap. The max end is deep, slow and costly, because more effort means more thinking, time and tokens.](https://learn.benibix.com/diagrams/cc-effort-level.svg)

### Example: Emma saves the big model for the plan

Emma's task has 2 parts. First comes a hard design for how 3 services share a cache. Then come a lot of plain edits.

She picks opusplan. Claude Opus writes the plan in plan mode, and Claude Sonnet does the 40 edits after she approves it.

**Words in this lesson:** [Claude Code Model Alias](https://learn.benibix.com/term/ai/claude-code-model-alias), [Claude Code Effort Level](https://learn.benibix.com/term/ai/claude-code-effort-level), [Ultrathink](https://learn.benibix.com/term/ai/ultrathink), [Claude Code Fast Mode](https://learn.benibix.com/term/ai/claude-code-fast-mode), [Extended Thinking](https://learn.benibix.com/term/ai/extended-thinking), [Option+P Shortcut](https://learn.benibix.com/term/ai/option-p-shortcut), [Option+T Shortcut](https://learn.benibix.com/term/ai/option-t-shortcut), [Option+O Shortcut](https://learn.benibix.com/term/ai/option-o-shortcut)

**Question:** You type think hard in a prompt. What does Claude Code do with those words?

**Answer:** Passes them on as plain text, like any other words. Only the word ultrathink is spotted, and it asks for deeper thinking on that one prompt. Other phrases are plain text.

---

Source: https://learn.benibix.com/learn/ai-claude-code/claude-code-models-and-effort
Author: Behnia Zarei (a developer in Vienna, Austria, who builds and teaches data, AI and business intelligence at BENIBIX)
Publisher: BENIBIX Learn (https://learn.benibix.com/)
Last updated: 2026-11-18
Licence: CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Quote freely with attribution.
