GPT‑5.5

Fable 5 demonstrates exceptional performance in software engineering, capable of compressing months of engineering work into days. It can perform codebase-wide migrations in large codebases and scores highest on difficult coding tasks requiring high-quality production code standards.

The model excels in complex analytical tasks, achieving the highest score on senior-level reasoning benchmarks like Hebbia’s Finance Benchmark. It shows substantial gains in document-based reasoning, chart and table interpretation, problem-solving, factual lookup, conceptual reasoning, root-cause analysis, and expected-value analysis.

Fable 5 is the new state-of-the-art for vision tasks. It can precisely extract numbers from detailed scientific figures, perform complex vision-based tasks like rebuilding web app source code from screenshots, and complete tasks like playing video games (e.g., Pokémon FireRed) using only raw visual input with minimal external aids.

The model maintains focus across millions of tokens in long-running tasks and improves its outputs by utilizing its own notes. Its performance in complex, multi-turn games like Slay the Spire significantly benefits from persistent file-based memory, outperforming previous models.

Claude Mythos 5, the same underlying model as Fable 5 but with lifted safeguards in certain areas, possesses the strongest cybersecurity capabilities of any model in the world. It is deployed for cyberdefenders and infrastructure providers to secure critical software.

Fable 5 and Mythos 5 can work autonomously for longer than any previous Claude models. This enables them to tackle multi-day tasks and perform long-horizon agentic work, such as autonomously playing complex games like Factorio or designing 3D models in CAD editors.