We Love Claude

Pick a model

Switch between Opus, Sonnet, Haiku, and extended thinking right from the composer.

Available on
Desktop Web

What it is

The model picker on the right of the chat composer lets you choose which Claude model answers. You can switch mid-conversation — the next turn uses whatever’s currently selected.

Where to find it

The current line-up

  • Fable 5 — the newest model; as of the 2026-07-12 web sweep it is the default shown in the composer (labelled “Fable 5 High”).
  • Opus 4.8 — most capable, for the hardest reasoning.
  • Sonnet 4.6 — efficient for everyday tasks.
  • Haiku 4.5 — fastest, best for quick answers.
  • Effort — submenu setting how hard Claude works on a prompt: Low (default), Medium, High, Max. Higher effort means more thorough responses that take longer and use your limits faster. A Thinking toggle in the same submenu lets Claude reason for more complex tasks. Replaces the old single Adaptive thinking toggle. See Extended thinking.
  • More models › — older versions (Opus 4.6, Opus 3, Sonnet 4.5) for legacy or comparison work.

How to use it

  1. Click the model name on the right side of the composer.
  2. Pick a model, then optionally set Effort (Low / Medium / High / Max) or flip the Thinking toggle.
  3. Send. The reply uses the selected model.

Good to know

  • There’s no “right” model — start with the default (now Fable 5) and switch to Haiku for speed or a Sonnet model when you’re doing lots of short turns.
  • Higher Effort trades latency for quality. Good for research, debugging, and multi-step reasoning; overkill for “what’s the weather in Paris”. The separate Thinking toggle was removed on 2026-07-29 — Effort is now the single control, defaulting to High.

What’s changed recently

  • 2026-07-29 — The desktop picker now lists four models with one-line descriptions: Fable 5 (“For your toughest challenges”), Opus 5 (“For complex tasks”), Sonnet 5 (“Most efficient for everyday tasks”) and Haiku 4.5 (“Fastest for quick answers”), plus Effort and a More models entry. Observed on build 1.24012.9.
  • 2026-07-29 — On claude.ai the picker now reads “Opus 5 High”, while the macOS app on build 1.24012.9 still reads “Opus 4.8 High” — web is ahead of desktop. The desktop list is Fable 5 / Opus 5 / Sonnet 5 / Haiku 4.5 plus an Effort control.
  • 2026-07-12 — The composer default on claude.ai is now Fable 5 (labelled “Fable 5 High”); Sonnet 5 has also appeared in the line-up (observed on desktop). Verified in the 2026-07-12 web sweep.
  • 2026-06-18 — Default model updated to Opus 4.8; the single Adaptive thinking toggle replaced by an Effort control (Low / Medium / High / Max) plus a separate Thinking toggle.

Related