The Horizon
719 hours — up from 60 hours a year ago; AI can now work autonomously for over 4 weeks
What is the current The Horizon?
The AI Task Horizon, measured by METR (Model Evaluation and Threat Research), tracks the duration of professional tasks that frontier AI models can perform autonomously — currently at 14 hours 30 minutes for the leading model (Claude Opus 4.6). This metric quantifies AI capability growth in terms directly relevant to workforce displacement: as the task horizon extends, the range of human jobs that AI can substitute for expands. Source: METR benchmark evaluations.
719 hours — AI models can now handle tasks lasting 4.3 weeks autonomously
METR time horizon measures duration of tasks frontier AI models can complete at 50% success rate. Doubling every 4.2 months since 2023. Latest frontier model: Claude Opus 4.6.
Explore Further
Is this happening to you?
Are AI tools now doing tasks you used to be paid for?
How has The Horizon changed over time?
| Period | Value | YoY Change |
|---|---|---|
| Feb 2026 | 718.8 | +658.40 |
| Dec 2025 | 352.2 | +313.40 |
| Nov 2025 | 293 | +254.20 |
| Aug 2025 | 203 | +182.70 |
| May 2025 | 101.2 | +94.20 |
| Apr 2025 | 119.7 | +116.50 |
| Feb 2025 | 60.4 | +56.80 |
| Dec 2024 | 38.8 | +34.80 |
| Oct 2024 | 20.5 | +16.50 |
| Sep 2024 | 20.3 | — |
| Jun 2024 | 11.4 | — |
| May 2024 | 7 | — |
Frequently Asked Questions
What is the AI Task Horizon?
The AI Task Horizon measures the duration of professional tasks that frontier AI models can perform autonomously, as evaluated by METR (Model Evaluation and Threat Research). The current leading capability is 14 hours 30 minutes — meaning AI can now sustain autonomous work on complex tasks for over half a day.
Why does the task horizon matter for jobs?
As the task horizon extends, the range of human jobs that AI can substitute for expands. Tasks requiring sustained focus over hours — writing reports, analyzing data, coding features, processing documents — increasingly fall within AI capability. This directly affects the employment prospects of knowledge workers.
Where does the AI Task Horizon data come from?
METR (Model Evaluation and Threat Research) conducts standardized benchmark evaluations of frontier AI models, measuring their ability to perform real-world professional tasks autonomously over extended durations. Results are published as new model capabilities are evaluated.
Quick poll
Is this affecting you or your household?
Discussion
Get the numbers when they move.
New data drops, indicator updates, and ADI score changes — delivered when it matters. No spam.
or Create an Account for full access
Loading comments…