Pick a model

What it is

Documentation update — 7 October 2026. The model guide now includes Sonnet 5.5 and Opus 5.5. Thinking is mandatory in chat for both, Fable 5.1 and Opus 5. Recommended effort is model-specific; Enterprise administrators may restrict choices or set starting defaults. The menu below was inspected on this account on desktop and web on 7 October; other accounts can differ.

The pill on the right of the chat composer picks which Claude model answers, and how much effort it puts in. Its label shows both at once — Fable 5.1 Medium, say. You can switch mid-conversation; the next turn uses whatever is selected.

Where to find it

On desktop

Click the model-and-effort pill below the unified composer, beside the project and permission controls.

On the web

The same model-and-effort pill sits below the composer on claude.ai.

Line-up inspected on 7 October

Four models, each with a one-line description in the menu:

Fable 5.1 — For your toughest challenges.
Opus 5.5 — For complex work and everyday tasks.
Sonnet 5.5 — Most efficient for simpler tasks.
Haiku 4.5 — Fastest for quick answers.

Below them sit two submenus:

Effort — five levels: Low, Medium (marked Recommended for the selected Opus 5.5), High, Extra, Max. The current level is shown in the row label. See Extended thinking.
More models — older releases, as a flat list with no descriptions: Fable 5, Opus 5, Opus 4.8, Opus 4.7, Opus 4.6, Opus 3, Sonnet 5, Sonnet 4.6. Both inspected platforms showed all eight.

How to use it

Click the model pill on the right of the composer.
Pick a model. Nothing is greyed out and nothing carries a Beta or New badge.
Optionally open Effort and pick a level, or More models for an older release.
Send. The reply uses whatever the pill now reads.

Good to know

Effort is a separate setting from the model. Anthropic’s article treats them as independent, and the effort selector covers Sonnet 5.5, Opus 5.5, Fable 5.1, Opus 5, Sonnet 5, Fable 5, Opus 4.7, Opus 4.6 and Sonnet 4.6.
The menu warns you about Fable. Alongside the model rows sits the line “Fable is the most capable model and draws down usage much faster than Opus.”
No separate Thinking toggle was visible with Opus 5.5 selected on 7 October. Anthropic’s article describes one, and also says thinking is mandatory on Sonnet 5.5, Opus 5.5, Fable 5.1 and Opus 5 in chat.
Enterprise administrators can configure the starting model and effort. You can still change the available selections within a conversation.
More models isn’t in version order. Opus 3 sits between Opus 4.6 and Sonnet 5.

What’s changed recently

2026-10-07 — Both inspected pickers now show Fable 5.1, Opus 5.5, Sonnet 5.5 and Haiku 4.5, plus eight More models entries. Effort marks Medium Recommended for the selected Opus 5.5; Max warns of 4× or more usage.
2026-10-07 — documentation review. Official model guide adds Sonnet 5.5 and Opus 5.5 and clarifies model-specific effort defaults and Enterprise controls. Both current picker menus were inspected on this account; other plan availability and response behavior were not tested.
2026-09-13 — Fable 5.1 has joined the top of the line-up, above Opus 5, Sonnet 5 and Haiku 4.5. More models now holds Fable 5, Opus 4.8, Opus 4.7, Opus 4.6, Opus 3 and Sonnet 4.6 — ten models in Chat mode across the two tiers. The four-model tier and Effort are the same on desktop and web in both composer modes; desktop drops Opus 3 from More models in Cowork mode, leaving nine. Verified on desktop build 1.52386.3.
2026-09-13 — Effort has five levels — Low, Medium (Default), High, Extra, Max — and the Default pill now sits on Medium, not High. The Thinking toggle the page previously described is not in the submenu.
2026-09-13 — Removed the walkthrough video. It was recorded on 4 May 2026, two line-up changes ago, and shows neither Fable 5.1 nor the five-level Effort submenu. A re-record is needed.
2026-07-29 — The desktop picker moved to four models with one-line descriptions, plus Effort and More models.