Anthropic's Claude 5 (Fable 5) dominates benchmarks, available only until June 22
Cutting-edge across SWE, vision, and science – but limited-time access sparks speculation.
A new model performs at the cutting edge across almost all major AI benchmarks, with especially strong results in software engineering, knowledge work, vision tasks, and scientific reasoning. It performs better on longer and more complex tasks, where it pulls ahead of other models, and makes better use of tokens than previous Claude versions. The model also stays consistently focused over extremely long contexts up to millions of tokens, and even improves its output by referring back to its own notes during long-running tasks.
- Leads nearly all major benchmarks, with standout performance in software engineering, vision, and scientific reasoning.
- Excels on long, complex tasks, maintaining focus over millions of tokens and using tokens more efficiently than prior Claude models.
- Available for a limited time until June 22, fueling speculation about Anthropic's intentions for the model's future.
Why It Matters
A brief window to test Anthropic's most powerful model—could redefine enterprise AI workflows but raises questions about release strategy.