Claude Fable 5 “Feels Next Level” via @sejournal, @martinibuster

Anthropic announced Claude Fable 5, its new generally available flagship AI model. The company says Fable 5 is its most capable publicly accessible model yet and is designed to handle longer, more complex tasks than previous Claude models. Anthropic highlighted improvements in coding and site development, knowledge work and research, vision and long-context tasks, new safety limits that affect how some requests are handled, and availability across the Claude platform and API.

Coding And Site Development

Jamie Marsland of Automattic gave Fable 5 a try and remarked that Fable 5 is “next level” after using it to generate a fully editable WordPress block theme.

Marsland tweeted:

“First test: can Fable 5 build a WordPress block theme?

One shot. Fully editable. Native WordPress patterns.

Yeah… this feels next level.”

Screenshot Of Test Site Created With Fable 5

Anthropic says Fable 5 delivers some of its strongest gains in software engineering, explaining that the new model can work autonomously for longer periods and perform complex coding tasks with less human oversight than previous Claude models.

As an example, Anthropic cited testing by Stripe, which reported that Fable 5 completed a codebase-wide migration in a 50-million-line Ruby codebase in a single day. Anthropic said the same task would have otherwise required a team working for more than two months.

The company also highlighted benchmark performance, noting that Fable 5 achieved the highest score among frontier models on Cognition’s FrontierCode evaluation, which measures performance on difficult coding tasks in production-style codebases.

Anthropic says Fable 5 can rebuild a web application’s source code from screenshots, a capability that combines software engineering and visual understanding.

Given what Marsland shared about Fable 5’s capabilities in a “one-shot” demonstration, Anthropic’s new model may represent a significant advance for real-world web development projects. For sites built with WordPress, Astro, and other modern frameworks, that could mean help with themes, blocks, templates, components, API connections, migrations, and debugging. The point is not just that Fable 5 can generate code, but that it appears capable of working across an entire project and making changes that depend on understanding how the pieces fit together.

Knowledge Work And Research

Anthropic also positions Fable 5 as a model for complex analytical and knowledge-work tasks.

The company says Fable 5 achieved the highest score on Hebbia’s Finance Benchmark for senior-level reasoning and showed improvements in document-based reasoning, chart interpretation, problem solving, and analysis. IMC reported strong performance across factual lookup, conceptual reasoning, root-cause analysis, and expected-value analysis.

These capabilities are relevant to work that requires analyzing large amounts of information, synthesizing findings from documents, interpreting data, and carrying out multi-step research tasks.

For SEOs, publishers, and site owners, that kind of work maps to tasks like analyzing search performance data, reviewing large content inventories, comparing documents, finding patterns across reports, and turning messy information into decision-ready data.

Vision And Long-Context Tasks

Anthropic describes Fable 5 as its strongest vision model to date. The company says it can extract precise information from complex scientific figures and perform visual tasks that previously required additional tooling or support systems.

The model also received upgrades in memory and long-context performance. Anthropic says Fable 5 can remain focused across millions of tokens and improve its work by referring to notes it has created during long-running tasks.

For users working with large collections of documents, lengthy projects, screenshots, images, and complex workflows, Fable 5’s long-context gains should prove useful for managing projects that are too complex for a simple prompt-and-answer exchange.

Safety Limits

Anthropic says Fable 5’s capabilities required new safeguards before a broad public release.

The company introduced classifiers that detect certain categories of requests and route them to Claude Opus 4.8 instead of allowing Fable 5 to answer directly. According to Anthropic, the affected categories include cybersecurity, biology and chemistry, and attempts to extract model capabilities through distillation.

Anthropic says the safeguards were intentionally configured conservatively to speed deployment while reducing misuse risks. The company acknowledged that some harmless requests may be caught by the system but said the safeguards are triggered in fewer than 5% of sessions on average.

The company explained:

“Without safeguards, Fable 5’s capabilities in areas like cybersecurity could be misused to cause serious damage.”

Anthropic says users will be informed whenever a request is routed to Opus 4.8.

Availability And Pricing

Claude Fable 5 is currently available on all paid plans beginning today through June 22nd, after which it will be available based on usage credits. The intention is to eventually make it available to paid plans in the future.

Anthropic explained:

“If capacity allows, we’ll extend the included window. After this point—when sufficient capacity allows us to do so—we aim to restore Fable 5 as a standard part of subscription plans. We intend to do this as quickly as we can.

Throughout this period, we’ll communicate any changes ahead of time so users know where things stand.”

For developers building applications with Claude, Anthropic has priced Fable 5 at $10 per million input tokens and $50 per million output tokens through the Claude API. Input tokens are the text, images, and other content sent to the model, while output tokens are the model’s generated responses.

Takeaway

Fable 5’s significance to developers, site owners, and SEOs lies in its ability to work across larger projects. Anthropic is presenting the model as a tool for coding, research, analysis, and long-running tasks that go beyond the simple prompt-and-response interactions associated with earlier AI systems.