Anthropic has released Claude Fable 5.1, a new generation of its AI model designed to handle complex coding, research and knowledge-work tasks with much less human supervision.
The model became available on September 1 and is being positioned by Anthropic as its most capable model for coding and knowledge work.
This Claude is designed to work for hours
One of the biggest changes with Fable 5.1 is its focus on long-running AI agents.
Instead of simply responding to individual prompts, the model can plan a task, use different tools, recover when something goes wrong and continue working through multiple stages.
Anthropic says Fable 5.1 can operate across applications, work through software backlogs, browse the web and handle complex projects with limited supervision.
That makes it particularly interesting for developers and companies looking to build AI agents that can actually complete projects rather than simply generate individual responses.
A serious upgrade for coding
Fable 5.1 is also aimed directly at software developers.
Anthropic says the model can work across entire codebases, perform code reviews, improve performance and run multi-day autonomous coding sessions.
It can also write its own tests and use vision to compare its output with the original design or requirements.
For developers, this points toward a future where AI assistants don't just write code — they can take responsibility for much larger portions of the software development process.
Anthropic cut the cost too
Perhaps one of the most interesting parts of the release is the pricing change.
Anthropic has reduced the price of cache reads by 75%, bringing them down to $0.25 per million tokens.
The company estimates that typical workloads can cost around 25% less than with Fable 5, while highly agentic workloads could see savings of up to approximately 45%.
Fable 5.1 itself is priced at $10 per million input tokens and $50 per million output tokens.
A 1-million-token context window
The model also comes with a 1-million-token context window, allowing it to work with extremely large codebases, documents and research projects.
That is particularly important for the type of long-running tasks Anthropic is targeting. Instead of repeatedly breaking a large project into smaller pieces, developers can give the model much more context and allow it to work through the problem over time.
Anthropic is also pushing AI research
Fable 5.1 isn't focused only on coding.
Anthropic says its latest models are also showing improved capabilities for scientific research, including tasks involving mathematics, science and multi-step research.
Alongside Fable 5.1, Anthropic introduced Claude Mythos 5.1, a version with the same underlying capabilities but additional safeguards for sensitive areas such as cybersecurity and life sciences. Mythos is currently available only through trusted-access programs.
Another major player in the AI race
The timing is significant.
OpenAI has GPT-6 Astra, Google has Gemini 3.8 Flash, DeepSeek has V4.1 Flash, and now Anthropic has Fable 5.1.
But the competition is increasingly shifting away from simple chatbot answers.
The next battle may be about which AI can actually take a complex project, work on it for hours and deliver a finished result with minimal human intervention.
Claude Fable 5.1 is clearly designed for that future.
