The length of task AI can complete is doubling every few months

The duration of software tasks (as measured by humans) that different models can complete 50% of the time

The plotted value is the duration at which a model has a 50% success rate. The exponential trend line for the plotted values is also shown. 95% confidence intervals are in the tooltips.
METR notes that measurements above 16 hours are unreliable with its current task suite, so the final point should be read with caution.
Chart: Centre for British ProgressSource: METRGet the data