Login or sign up
For business

Claude Fable 5

5.0· 3 reviews
Average ratingExcellent
5.0/ 5

Based on 3 reviews

5100%
40%
30%
20%
10%
Ask AI

Questions about Claude Fable 5

Based on 3 reviews, listing data and website checks · updated 3 days ago

What is Claude Fable 5?

Claude Fable 5 is an update for a model available on the Claude Platform, focused on long-running agent work, coding, and knowledge-heavy tasks. It features longer independent execution, staged planning, self-checking behavior, and safety classifiers that return errors for cybersecurity and biology requests. The public beta of Claude Managed Agents pairs Fable 5 with a harness for agentic work, with automatic fallback to Claude Opus 4.8.

What are the key technical changes in this update?

The main shift in Claude Fable 5 is longer independent execution, allowing it to plan across stages and check work along the way without prompt surgery. It is designed to work independently for longer than earlier generally available Claude models, with stronger staged planning and self-checking behavior. This reduces the need for constant hand-holding during multi-step tasks.

What are the current pricing and data retention policies?

Pricing for Claude Fable 5 is listed at $10 per million input tokens and $50 per million output tokens. Inputs and outputs are retained for 30 days to support safety work, which admins may want to consider when setting up separate workspaces. Blocked requests due to safety classifiers are not charged.

What do reviewers praise or complain about regarding this model?

Some users on Reddit say the model feels less like a standard launch and more like a preview of AI inequality, with heavy safety routing that may flag harmless requests. The system is described as a Mythos-class model that is state of the art on nearly all tested benchmarks, with exceptional performance in software engineering and scientific research. It can run for days, with performance scaling with task length.

How does the safety and fallback mechanism work?

Claude Fable 5 includes safety classifiers that return errors for cybersecurity and biology requests instead of generating model text. The automatic fallback feature allows blocked requests to continue on Claude Opus 4.8, and Managed Agents build this fallback automatically. This ensures that the vast majority of coding work remains unaffected by these safeguards.

What is the advisor strategy and how does it help teams?

The advisor strategy allows lower-cost worker models to call Fable 5 mid-task to check plans and evaluate work before shipping results. This gives cheaper workers a way to ask for high-signal review, while managed agents reduce the amount of custom harness work needed up front. The combination of longer task stamina and advisor patterns is considered useful for teams building agents.

What recent events or government directives have affected this model?

Some users on Reddit note that the model was abruptly disabled for all customers to comply with an urgent export control directive from the U.S. government. Following conversations with the government, Anthropic updated cybersecurity safeguards, which now flag a slightly higher fraction of harmless requests. Users are clearly notified when a request is flagged and receive a response instead of model text.

How is access to Claude Fable 5 structured for different subscription tiers?

Beginning July 20, Claude Fable 5 will be included in all Max and Team Premium plans at 50% of limits. Pro and Team Standard users will continue to have access via usage credits and will receive a one-time $100 credit. Demand for Fable has been challenging to predict, leading to a staged rollout to subscription plans.

What is the current status of the website and deployment?

The website mentions that the live domain for fable.pics was misrouting to Brisas Marinas during deployment. The build includes a dedicated page and vhost config needed to give Fable its own destination, though remote access to the VDS required a new login check. The static page is prepared locally with nginx config aligned to the launch text.

How does the model perform on extended tasks compared to previous versions?

The model is described as capable of running for days, with the longer the task, the larger the model's ability to handle it. It is designed to fit into agent harnesses without prompt surgery, making it suitable for hard coding tasks and knowledge-heavy flows. The operational read emphasizes that the useful part is the combination of longer task stamina, fallback behavior, and the advisor pattern.

How is Claude Fable 5 rated on Trustburn?

On Trustburn, Claude Fable 5 has a rating of 5 out of 5 based on 3 reviews.

AI-generated from reviews on this page. Verify anything important with the company directly.

Elena Rossi
Reviews 1

It is encouraging to see that Claude Fable 5 is back on the platform and available for longer-running agent work. The dedicated static site makes the documentation clear and accessible. Having a specific destination for the Fable domain improves the user experience significantly compared to previous misrouting issues.

David Miller
Reviews 1

From the information available, Claude Fable 5 presents itself as a robust solution for teams needing longer-running agent work. The ability to plan across stages and check work independently reduces the need for constant hand-holding during multi-step tasks. This independent execution capability seems particularly valuable for hard coding tasks and knowledge-heavy flows. The integration of automatic fallback to Claude Opus 4.8 for blocked requests adds a layer of reliability that supports continuous operation. For developers building agent harnesses, this model appears well-suited to fit into existing workflows without requiring significant prompt surgery, thereby streamlining the development process significantly.

News

No news yet. This is where the business shares its own updates and press.

Photos

No photos yet. Photos shared by reviewers will show up here.

Widgets

Boost your business trust by displaying a widget on your website — a simple way to convert your visitors into customers.

Get the widget for your site →

About Claude Fable 5

Claude Fable 5 is presented as an update note for a model available on the Claude Platform, focused on long-running agent work, coding, and knowledge-heavy tasks. The page describes new features such as longer independent execution, staged planning, self-checking behavior, and safety classifiers that return errors for cybersecurity and biology requests. It also mentions a public beta of Claude Managed Agents pairing Fable 5 with a harness for agentic work, automatic fallback to Claude Opus 4.8, and an advisor strategy where lower-cost worker models can call Fable 5 mid-task. Pricing is listed at $10 per million input tokens and $50 per million output tokens, with a 30-day data retention policy for inputs and outputs.

Website
fable.pics