Here is how to use Claude Fable: pick Claude Fable 5.1 in the model picker of the Claude app on a Pro, Max, Team or Enterprise plan, run /model fable in Claude Code, or call the model ID claude-fable-5-1 through the Claude API. Nothing extra to install beyond the app or tool you already use. The steps below cover each route as of October 2026, based on Anthropic's own pages.
If you would rather understand the model before trying it, ClaudeGPT Si, a free AI news site, explains each Claude release with sources listed under every story.
Before you start
Check three things first.
- Your plan. Claude's pricing table marks Fable as not available on the Free plan. Pro users reach it through usage credits; Max, Team and Enterprise plans include it, with Team access on premium seats.
- Data retention. By default, using Fable means 30-day data retention for safety monitoring.
- Safeguards. Requests in some areas, mostly offensive cybersecurity and dual-use biology, may switch to an Opus model.
Option 1: in the Claude app
- Open Claude on the web, desktop or mobile and start a new chat.
- Open the model picker and choose Claude Fable 5.1.
- Give it a big task. Fable is built for work that takes many steps, not quick questions.
Anthropic's launch post says Fable 5.1 defaults to Medium effort on Claude.ai and in Cowork. If a flagged request switches to Opus, the app shows a notice, and you can pick Fable again from the model picker. To stop automatic switching, go to Settings, then Capabilities, and turn off "Switch models when a message is flagged". With it off, a blocked request pauses so you can edit and retry.
Option 2: in Claude Code
Fable is not the default model on any plan, so you select it yourself.
| What you want | Command |
|---|---|
| Switch the current session to Fable 5.1 | /model fable |
| Start a new session on Fable 5.1 | claude --model fable |
| Use Fable 5 instead | /model claude-fable-5 |
Fable 5.1 needs Claude Code v2.1.257 or later; run claude update if a request fails. It defaults to High effort in Claude Code. The automatic switching setting lives under Config, then MODEL & OUTPUT.
Anthropic's tips for getting the most from it: describe the result you want rather than every step, skip reminders to test or check its work, and give it tasks you would usually split into pieces.
Option 3: through the API
Developers call claude-fable-5-1 on the Claude Platform, or the matching ID on Amazon Bedrock, Claude Platform on AWS, Google Cloud and Microsoft Foundry. The model has a 1M token context window and always-on adaptive thinking. One API change to know: the any and tool values for tool_choice return an error on Fable 5.1, so use auto with strict tool use instead.
When to use something else
Fable is the slowest of Anthropic's current models. For quick answers and everyday tasks, Anthropic suggests Claude Opus 5.5, which it says performs at the level of Fable 5.1 on most work. And Anthropic does not recommend Fable for professional biology research and drug development for now.
To follow what changes next, ClaudeGPT Si's Models section logs each Claude release with dates, and its Claude Code auto mode guide covers a feature that works with Fable models.
Frequently asked questions
How do I select Claude Fable in Claude Code?
Run /model fable in a session, or start one with claude --model fable. You need Claude Code v2.1.257 or later.
Can I use Claude Fable on the free plan? No. Claude's pricing table marks Fable as not available on Free.
Why did Claude switch from Fable to Opus mid-chat? A safety check flagged the request, so Claude re-ran it on an Opus model. You can switch back from the model picker.