Anthropic Claude Fable 5 on AWS: Mythos-class capabilities with built-in safeguards now obtainable

Up to date on June, 12, 2026 – Claude Fable 5 and Claude Mythos 5 on Amazon Bedrock entry unavailable

To assist compliance with the US Authorities export management directive, Anthropic has requested AWS to revoke entry to Claude Fable 5 and Claude Mythos 5 for all customers. All different fashions, together with Opus4.8, will not be affected and you’ll proceed utilizing them in full confidence. Please view the Anthropic assertion for additional particulars.

At present, we’re saying the provision of Claude Fable 5 on Amazon Bedrock and Claude Platform on AWS. Claude Fable 5 makes Mythos-level capabilities obtainable to prospects, with robust safeguards designed to make it secure for broader use. Fable 5 is state-of-the-art on almost all examined benchmarks and delivers distinctive efficiency in software program engineering, information work duties, and imaginative and prescient – constructed for formidable, lengthy working work.

With Claude Fable 5 on Bedrock, you may construct inside your current AWS surroundings and scale inference workloads. You can even use Claude Fable 5 by the Claude Platform on AWS, providing you with Anthropic’s native platform expertise.

In accordance with Anthropic, Claude Fable 5 represents a step-change in what you may accomplish with AI fashions. Here’s what makes this mannequin totally different:

• Lengthy-running, asynchronous execution — Claude Fable 5 handles advanced duties that earlier fashions couldn’t maintain, executing coding and information work duties for prolonged intervals with out intervention.
• Superior imaginative and prescient capabilities — Claude Fable 5 understands diagrams, charts, and tables nested in information and PDFs. This opens up analysis and document-heavy work in finance, authorized, analytics, structure, and gaming. In coding, the mannequin implements designs with excessive constancy and makes use of imaginative and prescient to critique its output towards targets.
• Proactive self-verification — The mannequin updates its personal abilities primarily based on learnings and develops its personal harnesses and evaluations.

Claude Fable 5 contains safeguards that restrict its efficiency in particular areas the place misuse threat is elevated. Dangerous prompts associated to cybersecurity, biology, chemistry, and well being fall again to obtain a response from Opus 4.8 as a substitute. Anthropic is ready to develop entry to just about all of Claude Fable 5’s state-of-the-art capabilities by creating extra highly effective safeguards. The identical mannequin with out these limits is Claude Mythos 5 and it’ll solely be obtainable to a small group of vetted prospects.

Claude Fable 5 mannequin in motion

You should use Claude Fable 5 in each Amazon Bedrock and Claude Platform on AWS. This put up covers steering on the best way to entry and use on Amazon Bedrock. For steering on the Claude Platform on AWS, go to the documentation to be taught extra.

To get began with Amazon Bedrock, you may entry the mannequin programmatically now utilizing the Anthropic Messages API to name the bedrock-runtime or bedrock-mantle endpoints by Anthropic SDK. You can even maintain utilizing the Invoke and Converse API on bedrock-runtime by the AWS Command Line Interface (AWS CLI) and AWS SDK.

Configure information retention setting

As a way to entry Claude Fable 5 mannequin, it’s essential to decide into information sharing by utilizing the Knowledge Retention API and setting provider_data_share earlier than you may invoke the fashions. There is no such thing as a console person interface for this setting at launch.

This mode permits Amazon Bedrock to retain and share your inference information with mannequin suppliers per their necessities. Anthropic requires 30-day inputs and outputs retention, in addition to human evaluate. To be taught extra, go to the Amazon Bedrock abuse detection.

Here’s a pattern script to set information retention for the bedrock-mantle engine.

curl -X PUT https://bedrock-mantle.us-east-1.api.aws/v1/data_retention 
  -H "x-api-key: "  
  -H "Content material-Sort: software/json" 
  -d '{ "mode": "provider_data_share" }'

If you wish to use the bedrock-runtime engine, run this pattern script.

curl -X PUT https://bedrock.us-east-1.amazonaws.com/data-retention 
  -H "Authorization: Bearer " 
  -H "Content material-Sort: software/json" 
  -d '{ "mode": "provider_data_share" }'

Up to date on Jun 10, 2026 — You can even use AWS SigV4 (Signature Model 4) to name the info retention API. Configure your AWS CLI or AWS SDK utilizing surroundings variables.

export AWS_ACCESS_KEY_ID=your_access_key_id
export AWS_SECRET_ACCESS_KEY=your_secret_access_key
export AWS_SESSION_TOKEN=your_session_token

First, retrieve your present Bedrock information retention settings.

curl -s https://bedrock.us-east-1.amazonaws.com/data-retention 
  --aws-sigv4 "aws:amz:us-east-1:bedrock" 
  --user "$AWS_ACCESS_KEY_ID:$AWS_SECRET_ACCESS_KEY" 
  -H "x-amz-security-token: $AWS_SESSION_TOKEN"

This could return one thing like this: {"mode":"inherit","updatedAt":null} and replace the info retention settings.

curl -s -X PUT https://bedrock.us-east-1.amazonaws.com/data-retention 
  --aws-sigv4 "aws:amz:us-east-1:bedrock" 
  --user "$AWS_ACCESS_KEY_ID:$AWS_SECRET_ACCESS_KEY" 
  -H "x-amz-security-token: $AWS_SESSION_TOKEN" 
  -H "Content material-Sort: software/json" 
  -d '{"mode":"provider_data_share"}'

If the whole lot labored as anticipated, it’s best to obtain a response like: {"mode":"provider_data_share","updatedAt":"2026-06-10T16:51:39.331Z"}.

The newest AWS CLI helps configuring the info retention setting. Set your bearer API key as an surroundings variable after you generate a API key within the Bedrock console.

export AWS_BEARER_TOKEN_BEDROCK=bedrock-api-key-XXXXXXXXXX

aws bedrock put-account-data-retention  
  --mode provider_data_share

To be taught extra, go to the Knowledge Retention API on the Amazon Bedrock Consumer Information.

use the Claude Fable 5 mannequin

Let’s begin with Anthropic SDK for Python utilizing the Messages API on bedrock-mantle endpoint. Set up Anthropic SDK.

pip set up anthropic

Here’s a pattern Python code to name Claude Fable 5 mannequin:

import anthropic

consumer = anthropic.Anthropic(
    base_url="https://bedrock-mantle.us-east-1.api.aws/anthropic",
    api_key= 
)

message = consumer.messages.create( 
     mannequin="anthropic.claude-fable-5", 
	 max_tokens=4096, 
	 messages=[ 
	     { "role": "user", 
		   "content": "Design a distributed architecture on AWS in Python that should support 100k requests per second across multiple geographic regions", 
		 }, 
	 ], 
)

print(message.content material[0].textual content)

To be taught extra, try Anthropic Messages API code examples and pocket book examples for a number of use circumstances and quite a lot of programming languages.

You should use Claude Fable 5 within the Bedrock console. Select Claude Fable 5 within the Playground and check it.

You can even use Claude Fable 5 with the Invoke API and Converse API on bedrock-runtime endpoint. Right here’s an instance to name Converse API for a unified multi-model expertise utilizing the AWS SDK for Python (Boto3):

import boto3 
bedrock_runtime = boto3.consumer("bedrock-runtime", region_name="us-east-1") 
response = bedrock_runtime.converse( 
    modelId="world.anthropic.claude-fable-5", 
    messages=[ 
        { 
            "role": "user", 
            "content": [ 
                { 
                    "text": "Design a distributed architecture on AWS in Python that should support 100k requests per second across multiple geographic regions." 
                } 
            ] 
        } 
    ], 
    inferenceConfig={ 
        "maxTokens": 4096 
    } 
) 
print(response["output"]["message"]["content"][0]["text"]) 

To be taught extra, go to code examples that present the best way to use Amazon Bedrock Runtime with AWS SDKs.

Issues to know

Let me share some vital technical particulars that I feel you’ll discover helpful.

• Mannequin entry — Claude Fable 5 entry is step by step increasing for all AWS accounts. In case your account doesn’t have entry but, it will likely be enabled quickly relying in your Bedrock utilization. If you wish to get entry to this mannequin shortly, contact your standard AWS Help.
• Pricing — When a dangerous immediate is routed to Opus 4.8 as a substitute of Fable 5, you pay solely Opus costs. If a request is blocked mid-conversation, preliminary tokens are charged at Fable charges and subsequent tokens at Opus charges. To be taught extra, go to the Amazon Bedrock pricing web page.
• Knowledge retention — For Fable 5, Mythos 5, and future fashions on Bedrock with related or greater functionality ranges, Anthropic would require 30-day retention for all visitors on Mythos-class fashions. Retaining information for a restricted interval permits Anthropic to detect patterns of misuse that aren’t seen from a single change. As soon as you choose into information retention, your information will go away AWS’s information and safety boundary.
• Claude Mythos 5 on Bedrock (Restricted Preview) — You can even use Anthropic’s most succesful mannequin for cybersecurity and life sciences, together with vulnerability discovery, drug design, and biodefense screening. Entry is at the moment restricted as a result of dual-use nature of those domains. To be taught extra, go to the mannequin card documentation.

Now obtainable

Anthropic’s Claude Fable 5 mannequin is on the market right now on Amazon Bedrock within the US East (N. Virginia) and Europe (Stockholm) Areas; test the full record of Areas for future updates. Claude Fable 5 can be obtainable on the Claude Platform on AWS in North America, South America, Europe, and Asia Pacific.

Give Claude Fable 5 a strive with the Amazon Bedrock APIs, within the Claude Platform on AWS, and ship suggestions to AWS re:Publish for Amazon Bedrock or by your standard AWS Help contacts.

— Channy

Up to date on June 9, 2026 — 1) Up to date the console screenshot. You should use the console on bedrock-runtime engine. The console assist on bedrock-mantle is coming quickly. 2) Fastened the suitable mannequin id within the pattern code, 3) Fastened right provider_data_share parameter, 4) Add a knowledge retention setting script for bedrock-runtime engine.

Up to date on June 10, 2026 —Added the best way to configure information retention setting by AWS SigV4 and AWS CLI.