Claude Fable 5

Software Engineering: Claude Fable 5 compresses months of engineering into days. It is state-of-the-art on frontier coding evaluations, capable of codebase-wide migrations, complex implementations, and multi-day autonomous sessions.

Agentic Workflows: Built for your most ambitious, long-running projects. Run Claude Fable 5 in an agent harness and it can work for days at a time: planning across stages, delegating to sub-agents, and rigorously checking its own work.

Vision and Knowledge Work: Fable 5 understands diagrams, charts, and tables nested in files and PDFs. It is the new state-of-the-art for vision tasks, capable of rebuilding web apps from screenshots alone.

Claude Fable 5 is the state-of-the-art model on CursorBench and FrontierBench. It opens up a class of long-horizon problems that were out of reach for earlier models. In a 50-million-line Ruby codebase, the model performed a codebase-wide migration in a day that would otherwise have taken a whole team over two months by hand.

Fable 5 is the new state-of-the-art model for tasks involving vision. It can extract precise numbers from detailed scientific figures and can perform complex vision-based tasks like rebuilding a web app's source code from screenshots alone.

Fable 5 stays focused across millions of tokens in long-running tasks and improves its outputs using its own notes. When given access to persistent file-based memory, its performance improves dramatically, allowing it to tackle days-long, complex, and asynchronous tasks.

Fable 5 includes robust safeguards for cybersecurity and biology. The model is tuned conservatively to ensure safety, utilizing classifiers to detect and block potentially dangerous requests, routing them to Opus 4.8 when necessary.