Get Zlio on your home screenOne tap, full screen, like an app.
Research · first seen 20 November 2026, 18:00 UTC
METR finds AI models can handle week-long human tasks
Time-horizon doubling is a key benchmark for how fast agentic capabilities advance.
Nonprofit METR reported that recent models including OpenAI's GPT-6 Astra can complete tasks a human would need about a full work week for, with capability time horizons doubling every 3.6 months, even as research judgment lags.