Claude Fable 5.2: How to Use It in Claude Code and the API
Find out how to use Claude Fable 5.2 in the Claude app, Claude Code, the API, OpenRouter and partner cloud platforms.
👁 114 reads
If you type /model fable into an old version of Claude Code, you will not get the new model. Claude Code’s documentation says Fable 5.2 requires version 2.1.321 or later. That is the first thing to watch out for if you want to try Claude Fable 5.2. The second is the Claude app, where Fable is not included in the Free plan. The third is the API, where old settings copied from other models can trigger a 400 error.
I checked the documentation, pricing pages and promotion terms for every route to Fable 5.2: the claude.ai app, Claude Code, the Anthropic API, OpenRouter and partner cloud platforms. Anthropic’s supported-countries list included Poland when I checked it, and the announcement did not mention regional restrictions for this model. Availability can vary by country, so check Anthropic’s current supported-countries list. The main differences come down to plans and settings.
Claude app: your plan determines access
Fable 5.2 is available on the Pro, Max, Ultra, Team and Enterprise plans. It is not included in the Free plan, and Anthropic’s help centre says the launch promotions do not apply to Free. On a paid plan, select it from the model menu, just as you would Opus or Sonnet.
How much you can use it depends on your plan. On Pro, which costs $20 per month, Fable is normally billed separately using usage credits. For launch, however, Anthropic offered a week of use at no extra cost, through November 16 at 11:59 p.m. Pacific Time, with a limit of half the weekly allocation and one free reset. On Max ($100) and Ultra ($200), Fable is included up to half the weekly limit. Subscribers to those plans received between one and three resets, depending on how long they had been subscribed. The resets are valid through March 8, 2027.
You trigger a reset yourself: Settings > Usage > Limit resets > “Reset for free”. It renews the weekly Fable limit and, if your account has one, the five-hour limit. It does not increase the limit, so save it for a day when you are likely to run out. Keep in mind that Fable uses up limits faster than other models. The default effort level in Claude apps is Medium.
Claude Code: three commands to know
In Claude Code, Fable 5.2 is not the default on any plan, so you need to select it. After updating to version 2.1.321 or later, you have two options:
/model fable
claude --model fableThe first switches models during a session; the second starts a session with Fable. The fable alias points to Fable 5.2 everywhere except sessions through the Claude apps gateway. There, it points to the older Fable 5 by default, so use the full identifier in those sessions: /model claude-fable-5-2.
Set the effort level with /effort, followed by one of these options: low, medium, high, xhigh or max. You can also set it for a single session at launch: claude --effort high. Fable 5.2 defaults to medium in Claude Code. According to Anthropic’s announcement, at medium effort the model uses fewer tokens than Fable 5.1 on difficult tasks; the best results in the benchmark table came at the highest effort levels.
The third command is /fast, which enables fast mode. Press Space to toggle it, then Enter to confirm; a ↯ icon appears by the message field. Fable generates text up to 2.5 times faster in this mode, but costs $16 per million input tokens and $80 per million output tokens, instead of $8 and $40. On subscription plans, fast mode uses usage credits only, even if you still have quota remaining, and you must enable credits in settings first. On Team and Enterprise, an organization owner must also enable it. The documentation recommends turning it on at the start of a conversation, because the first time you enable it, you pay the full fast-mode rate for all the context already in the conversation.
API access: the model ID and two changes to your workflow
In the Anthropic API, the model ID is claude-fable-5-2. Here is a basic request from the terminal. The message is an English translation of the Polish text used in the original example; it is not a newly tested prompt.
curl https://api.anthropic.com/v1/messages \
-H "x-api-key: $ANTHROPIC_API_KEY" \
-H "anthropic-version: 2023-06-01" \
-H "content-type: application/json" \
-d '{"model": "claude-fable-5-2", "max_tokens": 16000,
"messages": [{"role": "user", "content": "Summarize this report in 5 bullet points: ..."}]}'You can do the same in Python with Anthropic’s official anthropic library. The message text below is an English translation of the Polish text used in the original example, not a newly tested prompt.
import anthropic
client = anthropic.Anthropic() # Reads the key from ANTHROPIC_API_KEY
msg = client.messages.create(
model="claude-fable-5-2",
max_tokens=16000,
messages=[{"role": "user", "content": "Summarize this report in 5 bullet points: ..."}],
)
print("".join(b.text for b in msg.content if b.type == "text"))The last line selects only text blocks because Fable 5.2 always thinks before answering. Adaptive thinking cannot be turned off, and you control reasoning depth with the effort parameter, which defaults to high in the API. In Claude apps and Claude Code, the default is medium. Two things can break older scripts: Fable 5.2 returns a 400 error when tool use is forced with tool_choice set to any or tool, and the documentation says to omit temperature, top_p and top_k. The context window is one million tokens, the response limit is 128,000 tokens, and pricing is $8 per million input tokens and $40 per million output tokens.
One more point for businesses: Fable 5.2 requires 30-day data retention. It does not support zero data retention unless Anthropic explicitly approves it. Customers who need that have a separate option called Enterprise Frontier Safeguards.
My test results
I ran my Fable 5.2 test through OpenRouter, using medium effort for all 27 tasks. The test was conducted in Polish; any translated examples are translations of the original Polish test material, not new English-language tests. The median time to first token was 2.24 seconds, and a 500-word essay was ready in 19.1 seconds, 27% faster than with Fable 5.1. The complete set cost $0.75. One thing to keep in mind: in one of the three tasks that asked for clean JSON, Fable added calculations, so parse its output carefully.
Beyond Anthropic: OpenRouter and cloud platforms
If you would rather not create an account in the Anthropic console, you can use OpenRouter. When I checked its public model list on the evening of November 9, 2026, it included anthropic/claude-fable-5.2 at the same price as Anthropic: $8 per million input tokens and $40 per million output tokens, with a one-million-token context window. You can use it by changing the model name in code that already uses OpenRouter.
On partner cloud platforms, the model ID is the same: claude-fable-5-2 on Google Cloud, both in the Google-managed and Anthropic-managed versions, Microsoft Foundry and Claude Platform on AWS. Amazon Bedrock is the exception: the model table in its documentation shows a dash instead of an ID for Fable 5.2, although Opus 5.5 and Sonnet 5.5 have IDs there. For now, AWS users should look at Claude Platform on AWS. Fast mode in the API is available through Anthropic, Google Cloud and Anthropic-hosted Foundry; in Claude Code, it is available only through the Anthropic API or a subscription.
Anthropic offers a useful note of caution: if you are not sure which model to use, start with Opus 5.5, and turn to Fable 5.2 for difficult reasoning and long-running agent tasks. I have collected prices, benchmarks and model differences in this Claude Fable 5.2 guide.
Frequently asked questions
Is Claude Fable 5.2 available in my country?
Availability depends on the country and provider. Poland was on Anthropic’s supported-countries list when I checked, and the announcement did not mention regional restrictions. Fable 5.2 is available through paid Claude app plans, Claude Code, the API, OpenRouter and Google and Microsoft cloud platforms. Check the provider’s current availability information for your location.
How do I enable Fable 5.2 in Claude Code?
Update Claude Code to version 2.1.321 or later, then enter /model fable or launch it with claude --model fable. In sessions through the Claude apps gateway, use the full name: /model claude-fable-5-2.
Is Fable 5.2 included in Claude’s free plan?
No. Fable 5.2 is available from the Pro plan upwards. After the launch promotion ended on November 16, 2026, Pro users paid for it using usage credits.
What is the Fable 5.2 API model ID?
Use claude-fable-5-2 with the Anthropic API, Google Cloud, Microsoft Foundry and Claude Platform on AWS. On OpenRouter, use anthropic/claude-fable-5.2.
Sources
- Claude Code documentation: Model configuration (checked November 9, 2026)
- Claude Code documentation: Fast mode (checked November 9, 2026)
- Claude documentation: Claude Fable 5.2 (checked November 9, 2026)
- Claude documentation: Models overview (checked November 9, 2026)
- Claude documentation: Pricing (checked November 9, 2026)
- Claude help centre: Fable 5.2 launch promotion
- Claude: Plan pricing (checked November 9, 2026)
- Anthropic: Supported countries
- Anthropic: Introducing Claude Fable 5.2 and Claude Mythos 5.2 (November 9, 2026)
- OpenRouter model list (API /api/v1/models), checked on the evening of November 9, 2026
