AI model · Anthropic

Claude Fable 5.2, Explained: Specs, Pricing, Access

Q: What model ID does the API use for Fable?

On the Claude API the model ID and alias are both claude-fable-5-2. On Amazon Bedrock it is anthropic.claude-fable-5-2. Google Cloud, Microsoft Foundry and Claude Platform on AWS use claude-fable-5-2. Anthropic says the model will not be retired before November 9, 2027, so you can pin it in production for at least a year.

Claude Fable 5.2 is Anthropic's model for the hardest reasoning and long agent runs, released November 9, 2026 at $8 per million input tokens. Synexa does not serve it; pair it with Synexa when your app also needs images, video, speech or 3D from one API key.

1M-token context, 128K output
$8 in / $40 out per MTok
Synexa: image, video, audio, 3D APIs

Get Started

Nov 9, 2026

Release date, Anthropic's newest top model

Context window in tokens

128K

Maximum output tokens per response

$8 / $40

Input / output price per million tokens

What is Claude Fable 5.2?

Fable 5.2 is a large language model from Anthropic, released on November 9, 2026, and positioned for demanding reasoning and long-horizon agentic work: agent sessions that run for hours, multistep research and analysis carried through to a finished document, spreadsheet or deck. It replaces Claude Fable 5.1 at the top of Anthropic's generally available lineup.

Anthropic Fable 5.2 shares its underlying model with Claude Mythos 5.2. The difference is the safeguards. Fable ships with strong cybersecurity and biology protections, which is why anyone can use it, while Mythos goes to organizations approved through Anthropic's Cyber and Life Sciences verification programs. Anthropic notes that the Fable safeguards will block most cybersecurity tasks.

Anthropic does not suggest the new model as a default. Its guidance is to start most coding and knowledge work on Claude Opus 5.5 and move up to Fable when the task is open-ended, long, or still failing at higher effort.

Fable 5.2 specs: context, output, IDs and limits

Published figures as of November 2026.

Maker: Anthropic
Released: November 9, 2026
Claude API ID: claude-fable-5-2
Amazon Bedrock ID: anthropic.claude-fable-5-2
Input / output: Text and images in, text out
Context window: 1M tokens
Max output: 128K tokens
Thinking: Adaptive, always on; depth set by the effort parameter
Default effort (API): high
Knowledge cutoff: June 2026
Retirement: Not sooner than November 9, 2027
Data retention: Requires 30-day retention; zero data retention only if Anthropic authorizes it

How much does Fable 5.2 cost?

The model costs $8 per million input tokens and $40 per million output tokens on the Claude API, 20% less than Fable 5.1. Prices as of November 2026.

Input, per MTok

Fable 5.2: $8
Fable 5.1: $10
Opus 5.5: $4

Output, per MTok

Fable 5.2: $40
Fable 5.1: $50
Opus 5.5: $20

5-minute cache write

Fable 5.2: $10
Fable 5.1: $12.50
Opus 5.5: $5

1-hour cache write

Fable 5.2: $16
Fable 5.1: $20
Opus 5.5: $8

Cache read

Fable 5.2: $0.20
Fable 5.1: $0.25
Opus 5.5: $0.20

Batch API (50% off)

Fable 5.2: $4 / $20
Fable 5.1: $5 / $25
Opus 5.5: $2 / $10

Fast mode

Fable 5.2: $16 / $80
Fable 5.1: not listed
Opus 5.5: $8 / $40

Example: 100K in + 10K out

Fable 5.2: $1.20
Fable 5.1: $1.50
Opus 5.5: $0.60

Rate	Fable 5.2	Fable 5.1	Opus 5.5
Input, per MTok	$8	$10	$4
Output, per MTok	$40	$50	$20
5-minute cache write	$10	$12.50	$5
1-hour cache write	$16	$20	$8
Cache read	$0.20	$0.25	$0.20
Batch API (50% off)	$4 / $20	$5 / $25	$2 / $10
Fast mode	$16 / $80	not listed	$8 / $40
Example: 100K in + 10K out	$1.20	$1.50	$0.60

The example row is our arithmetic from the list prices: 0.1M × $8 plus 0.01M × $40 = $1.20 for Fable 5.2. The same call through the Batch API is $0.60.

What is new in Anthropic Fable 5.2?

The upgrade from Fable 5.1 is mostly about speed, price and clearer output.

Faster output

Anthropic reports output up to 30% faster than Fable 5.1, plus an optional fast mode that speeds up token generation by up to 2.5 times at $16 / $80 per million tokens.

Lower price

List price dropped from $10 / $50 to $8 / $40 per million tokens. Cache reads fell from $0.25 to $0.20, and batch from $5 / $25 to $4 / $20.

Fewer tokens at medium effort

On medium effort, the default in the Claude apps and Claude Code, Anthropic says the model finishes hard tasks with fewer tokens than Fable 5.1.

Fewer unnecessary blocks

Cyber and biology safeguards are described as just as strict as on 5.1, but they produce about 55% fewer blocks than Fable 5.1 in early testing. Blocked queries get an answer from an older Opus model instead.

Advisor and orchestrator roles

It can act as an advisor that Haiku, Sonnet or Opus consult on hard problems, or as an orchestrator that hands simpler subtasks to smaller models.

Better vision

Anthropic lists coding, reasoning and vision as the main areas of improvement, including chart reading.

How does the new Fable score on benchmarks?

These are scores Anthropic published at launch, with safeguards on and at each model's highest available effort unless noted. We have not run them ourselves.

FrontierSWE v2 (agentic coding)

Fable 5.2: 69.9%
Fable 5.1: 56.3%
Opus 5.5: 62.3%

FrontierCode 2.0 (agentic coding)

Fable 5.2: 49.1% at xhigh
Fable 5.1: 32.4% at xhigh
Opus 5.5: 45.5%

OSWorld 2.1 (computer use, partial set)

Fable 5.2: 89.3%
Fable 5.1: 83.5%
Opus 5.5: 87.2%

Humanity's Last Exam (with tools)

Fable 5.2: 74.2%
Fable 5.1: 65.6%
Opus 5.5: 67.7%

Remote Labor Index

Fable 5.2: 41.3% at high
Fable 5.1: 17.9%
Opus 5.5: 21.3%

Benchmark	Fable 5.2	Fable 5.1	Opus 5.5
FrontierSWE v2 (agentic coding)	69.9%	56.3%	62.3%
FrontierCode 2.0 (agentic coding)	49.1% at xhigh	32.4% at xhigh	45.5%
OSWorld 2.1 (computer use, partial set)	89.3%	83.5%	87.2%
Humanity's Last Exam (with tools)	74.2%	65.6%	67.7%
Remote Labor Index	41.3% at high	17.9%	21.3%

Vendor-reported numbers on Anthropic's own test setups. Treat them as a direction, then run your own evals.

How to access Fable 5.2

There are three routes: the Claude apps, the Claude API, and the big cloud platforms.

Claude apps

Free accounts can't pick Fable. Pro ($20 a month, or $17 a month billed yearly) pays for Fable with usage credits, though during launch week Pro also gets Fable for up to half of its weekly limit. Max ($100) and Ultra ($200) include it for up to half of the weekly limit. During launch week, eligible paid plans get one to three Fable usage resets.

Claude API

Call the Messages API with the model ID claude-fable-5-2, billed per token at the rates above. Batch requests get 50% off; fast mode costs double the standard rate.

Cloud platforms

The same model runs on Amazon Bedrock (anthropic.claude-fable-5-2), Claude Platform on AWS, Claude models on Google Cloud (Google-managed and Anthropic-managed) and Microsoft Foundry. Each cloud bills through its own account.

API gotchas to know before you migrate

A few request settings that worked on older Claude models now return errors.

Set thinking depth with the effort parameter: low, medium, high (default), xhigh or max
Use the Batch API for offline jobs to halve the bill
Cache long system prompts; reads cost $0.20 per million tokens
Start on Opus 5.5 and escalate the hard cases

Don't

Force tool use: tool_choice set to any or tool returns a 400 error
Send temperature, top_p or top_k; leave sampling parameters out
Try to turn thinking off; adaptive thinking is always on
Plan on zero data retention without written approval from Anthropic

Does Synexa run Fable 5.2?

No. Synexa does not serve this model or any other Claude model, and it offers no chat LLMs such as Claude or GPT; its only text outputs are LLaVA-13B image answers and Whisper speech transcripts. For the language side, use the Claude API or one of the clouds above.

Synexa covers the other half of most AI products: generative media through one REST API. One key gives you image models such as Nano Banana Pro at $0.10 a request, video, speech, upscaling, and image-to-3D models such as Hunyuan3D 3.1 at $0.16 a request. A common pattern is to have the language model write the prompt, then send it to Synexa to make the asset. Every price is listed on its model page in Explore, and new accounts start with $0.10 of free credit. Create a key on the API keys page and make your first call with the quickstart.

Pay per request, no subscription
Python and Node clients, plus plain HTTP
Image, video, audio, upscaling and 3D in one catalog

Is Fable 5.2 worth it over Opus 5.5?

Only for the tasks Opus 5.5 can't finish. Fable 5.2 costs twice as much per token as Opus 5.5 ($8 / $40 against $4 / $20), and Anthropic's own advice is to start on Opus. Anthropic also reports that on FrontierCode 2.0, the new model at high effort scores about level with Opus 5.5 at max effort, for roughly half the cost per task. So the upgrade can pay off on long agent runs where Opus burns tokens.

Good candidates for Fable: multi-hour coding migrations, research that ends in a written report, and agent systems where the model plans and smaller models execute. Poor candidates: chat, classification, extraction and routing, where Sonnet 5.5 or Haiku 5.5 is faster and far cheaper. If your app also produces images, voice or 3D assets, see how a dedicated media API compares in our WaveSpeed AI alternative page, or read about another new model, Tencent Hy4.

More model explainers and API comparisons.

Tencent Hy4

Tencent's newest flagship language model: preview and final releases, specs, price and how to run it.

Synexa vs WaveSpeed

Synexa compared with WaveSpeedAI: per-model prices, free credits, catalog and when each one fits.

Explore Synexa models

Browse the image, video, audio and 3D models you can run with one API key.

Synexa pricing

Per-request prices for the models on Synexa, with no subscription.

Frequently asked questions

What is Fable 5.2?

Fable 5.2 is Anthropic's model for demanding reasoning and long-horizon agent work, released November 9, 2026. It reads text and images, writes text, holds a 1M-token context and returns up to 128K tokens per response. It shares its underlying model with Claude Mythos 5.2 but ships with stronger cybersecurity and biology safeguards, which lets Anthropic offer it to everyone.

How much does Claude Fable 5.2 cost?

On the Claude API it costs $8 per million input tokens and $40 per million output tokens as of November 2026, down from $10 and $50 on Fable 5.1. The Batch API halves that to $4 and $20, cache reads cost $0.20 per million tokens, and the optional fast mode costs $16 and $80. A call with 100K tokens in and 10K out costs about $1.20.

Is Anthropic Fable 5.2 free?

No. The Claude Free plan doesn't include Fable models. Pro subscribers pay for Fable with usage credits, while the Max ($100 a month) and Ultra ($200 a month) plans include it for up to half of the weekly usage limit. On the API it is billed per token with no free tier, so you need a funded Claude Platform or cloud account.

What is the difference between Claude Fable 5.2 and Mythos 5.2?

They are the same underlying model. Fable adds strong cybersecurity and biology safeguards and is open to all users, but those safeguards block most cybersecurity tasks. Mythos 5.2 has reduced safeguards for vetted work and is only available to organizations accepted into Anthropic's Cyber or Life Sciences verification programs, or through Claude Security on the Enterprise plan.

What model ID does the API use for Fable?

On the Claude API the model ID and alias are both claude-fable-5-2. On Amazon Bedrock it is anthropic.claude-fable-5-2. Google Cloud, Microsoft Foundry and Claude Platform on AWS use claude-fable-5-2. Anthropic says the model will not be retired before November 9, 2027, so you can pin it in production for at least a year.

Why does my Fable 5.2 request return a 400 error?

The usual causes are settings that older Claude models accepted. The new model rejects forced tool use (tool_choice of any or tool), so let the model choose tools instead. It also rejects certain temperature, top_p and top_k values, so leave those parameters out. Thinking can't be turned off, so control depth with the effort parameter instead of disabling thinking.

Can I use Fable 5.2 through Synexa?

No. Synexa does not serve any Claude model or any other chat language model. Synexa is a pay-per-request API for generative media: images, video, speech and audio, upscaling and image-to-3D. Many teams call Claude for planning and writing, then call Synexa to render the image, clip, voice-over or 3D model. New Synexa accounts get $0.10 of free credit.

Should I upgrade from Fable 5.1 to Fable 5.2?

For most workloads, yes. Fable 5.2 is 20% cheaper per token, Anthropic reports output up to 30% faster, and it beat Fable 5.1 on every Fable-to-Fable row of the launch table. Before you switch, check that your requests don't force tool use or send sampling parameters, because those return errors. Then rerun your own evals on a sample of real tasks.

Pair Fable 5.2 with a media API

Use Fable 5.2 for the reasoning, then generate the images, video, speech and 3D models it plans on Synexa. Pay per request, start with $0.10 free credit.