Introducing ci-proto.

ci-proto is our first time series foundation model. It combines frontier forecasting accuracy with a context window of over two million steps, more than 60 times that of any other time series foundation model.

Read the launch post

Start forecasting with ci-proto.

pip install continuousintelligence

Run locally on a CPU or GPU.

Built for your time series

Use earlier events to forecast what comes next.

How will a household’s electricity use change during a scheduled power outage? In this experiment, ci-proto is given the outage schedule and a history of electricity use, including a similar scheduled power outage almost four years earlier.

With a long enough context, the model can use that earlier event to forecast the drop in usage. Switch between the two histories to see the effect.

Electricity use during a scheduled power outage

Simulated data

With 4 years of history, the earlier scheduled power outage is in context. The forecast captures the drop in electricity use.

One eight-hour window from the launch post’s experiment, using the ci-proto long-context checkpoint. Both forecasts receive the same future power outage schedule.

Is your organisation interested in working directly with our team to apply our unreleased, frontier time series solutions to your own data and problems?

Become a partner