Anthropic: Claude Fable 5.2

Latest Claude Fable model for demanding reasoning and long-horizon agentic work, with always-on adaptive thinking and fewer safety-classifier blocks than Fable 5.1

Compare models

claude-fable-5-2

AvailableAnthropicChatCached input 98% lessNew

Input / Output-65%: $8.00 / $40.00$2.80 / $14.00
Context: 1M
Released: Nov 9, 2026
Max output: 128K
Modalities: VisionChat
Capabilities: Tool usePrompt CacheReasoning

About Claude Fable 5.2

Claude Fable 5.2 is Anthropic's Claude Fable model released in November 2026 for demanding reasoning and long-horizon agentic work. It is the same model as Mythos 5.2 with cybersecurity and biology safeguards. Against Fable 5.1, Anthropic reports higher scores on coding, agentic, science and knowledge-work benchmarks, output up to 30% faster, and about 55% fewer safeguard blocks.

Where it works well

Anthropic reports large gains over Fable 5.1 on FrontierCode 2.0, FrontierSWE v2 and AutomationBench.
Output is generated up to 30% faster than Fable 5.1, according to Anthropic's launch post.
Anthropic says it finishes challenging tasks with fewer tokens than Fable 5.1 at medium effort.
Safeguards block about 55% fewer requests than Fable 5.1 in Anthropic's early testing, with the largest drop in cybersecurity.

When to choose another model

Anthropic suggests starting with Opus 5.5 and moving to Fable 5.2 only when evaluations at higher effort still fall short.
Thinking is always on, and manual or disabled thinking returns an error, so short replies still carry thinking overhead.
Classifiers still block exploit generation, penetration testing and most research biology, and route those requests to another Claude model.

Getting started

Create API key
Create a key in Console, then use it with every model on the platform.

Send your first request

Copy the example for your language and run it against the endpoint.

Text to textMessages API

POST/v1/messages

API format:

curl https://api.tokenlab.sh/v1/messages \
  -H "Content-Type: application/json" \
  -H "x-api-key: sk-xxx" \
  -H "anthropic-version: 2023-06-01" \
  -d '{
    "model": "claude-fable-5-2",
    "max_tokens": 1024,
    "messages": [
      {"role": "user", "content": "Hello!"}
    ]
  }'

Pricing

The Verified price applies to Verified, which costs less on most models. The Official price is the model maker's published price and applies to the more reliable Official route. Auto bills the route that completes the request.

Default spec

per 1M tokens

Official price: Input $8.00 / Output $40.00 / Cache read $0.20 / Cache write $10.00 / Cache write (1h) $16.00
Verified price: Input $2.80 / Output $14.00 / Cache read $0.07 / Cache write $3.50 / Cache write (1h) $5.60
Discount: -65%

Prompt cache pricing

Cache read

Official price: $0.20
Verified price: $0.07
Discount: -65%

Cache write

Official price: $10.00
Verified price: $3.50
Discount: -65%

Cache write (1h)

Official price: $16.00
Verified price: $5.60
Discount: -65%

Tool fees

Charged per call, only when the model uses the tool.

Web search

Official price: $0.01/search
Verified price: $0.0035/search
Discount: -65%

	Official priceper 1M tokens	Verified priceper 1M tokens	Discount
Default spec	Input $8.00 / Output $40.00 / Cache read $0.20 / Cache write $10.00 / Cache write (1h) $16.00	Input $2.80 / Output $14.00 / Cache read $0.07 / Cache write $3.50 / Cache write (1h) $5.60	-65%
Prompt cache pricing
Cache read	$0.20	$0.07	-65%
Cache write	$10.00	$3.50	-65%
Cache write (1h)	$16.00	$5.60	-65%
Tool feesCharged per call, only when the model uses the tool.
Web search	$0.01/search	$0.0035/search	-65%

Usage & activity

Success rate is the share of requests that completed. Latency is how long a full response takes; P95 means 95% of requests finished within that time.

Usage & availability

Last 24 hours

Requests
Success rate
P95 latency
Total tokens

Model performance

Last updated Nov 10, 2026, 11:49 PM UTC

Metrics appear once privacy and data volume thresholds are met.

Data is based on aggregate user requests, excluding status checks.

Open in Console

Open Claude Fable 5.2 in Console with a prompt ready to edit or send.

Help me try claude-fable-5-2 with a short message at /v1/messages. Show the reply, latency, and cost.

Use cases

Best for

Reasoning
Vision

Multi-hour coding agents
Run agent sessions that plan, write and test changes across a large codebase for hours, or coordinate many subagents on one migration.
Deep research to a finished file
Carry multistep research through to a finished document, spreadsheet or deck, with sources checked along the way.
Scientific and engineering analysis
Work through physics, materials or hardware design problems where long chains of reasoning and tool use decide the result.
Orchestrating smaller models
Use it as the planner or advisor that hands routine subtasks to faster Claude models and reviews their output.

Prompt examples

Migrate this service from Python to Rust, keep the public API unchanged, and run the test suite after each module.

Research how these three suppliers price their contracts, reconcile the figures, and deliver a spreadsheet with a summary sheet.

Review this financial model for errors, explain each one, and return a corrected version with the changes highlighted.

This model has conditional pricing. Monthly token totals alone cannot produce a reliable estimate; use the detailed pricing for the request specification, cache, and applicable time window.

FAQ

What is Claude Fable 5.2 best for?

Demanding reasoning and long-horizon agentic work: agent sessions that run for hours, multistep deep research, and analysis carried through to a finished document, spreadsheet or deck. Anthropic recommends it when evaluations on Opus 5.5 at higher effort still fall short.

How does Fable 5.2 differ from Fable 5.1?

Anthropic reports higher scores on every benchmark it compared, including FrontierCode 2.0, FrontierSWE v2, Humanity's Last Exam with tools and OSWorld 2.1. Output is up to 30% faster, medium effort needs fewer tokens, and the safeguards block about 55% fewer requests.

How is Fable 5.2 related to Mythos 5.2?

Anthropic states they are the same underlying model. Fable 5.2 adds safeguards that block most high-risk cybersecurity tasks and most research biology. Mythos 5.2 is limited to organizations verified through Anthropic's Cyber or Life Sciences Verification Programs.

Can I turn thinking off on Fable 5.2?

No. Adaptive thinking is always on, as on Fable 5.1, and manual or disabled thinking returns an error. Control depth with the effort parameter. Anthropic notes that prompt lines asking for less thinking do not reduce it at medium effort and above; lower the effort instead.

What changes when I move from Fable 5.1 to Fable 5.2?

Anthropic's migration guide lists a prefix check on every account: editing earlier turns invalidates thinking blocks. Thinking blocks are also bound to the account that produced them, and a very small max_tokens can end with no text after the thinking block.

How much does Claude Fable 5.2 cost?

On TokenLab, Claude Fable 5.2 costs Input $2.80 / Output $14.00 per 1M tokens. The pricing table above shows the full breakdown. Rates depend on the billing unit, specification, and usage. Compare matching conditions in the model's detailed pricing; a single rate does not determine the total cost.

What are the context window and output limit of Claude Fable 5.2?

Claude Fable 5.2 accepts up to 1,000,000 tokens of context and returns up to 128,000 tokens in one response.

Which endpoint should Claude Fable 5.2 use?

Use https://api.tokenlab.sh/v1/messages for Claude Fable 5.2. The request example below shows the matching code shape.

Which operations does Claude Fable 5.2 support?

Claude Fable 5.2 supports Text to text. Select an operation above to see its endpoint and request example.