Learn about this first
Epoch AI released Epoch Automation Reports, evaluating AI against the standards of human employees on the organization’s routine tasks, such as designing charts and planning research projects. The results show that leading models Claude Fable 5.1 and GPT-6 Astra score an average of 65%, below the 100% threshold for automation.
This evaluation focuses on Epoch AI’s own work tasks. The scores reflect how the models perform on these tasks against human standards, not a ranking of general capabilities.
AI-generated from reports · updated 21 days agoLatest developments
Epoch AI evaluates frontier models on internal work tasks: Claude Fable 5.1 and GPT-6 Astra average only 65%
Event progress1
Epoch AI releases an automation evaluation report
Epoch AI: Gradient Updates: Epoch AI evaluates frontier models on internal work tasks: Claude Fable 5.1 and GPT-6 Astra average just 65%