The forecast competition leaves its data open.

The human problem

Communities and weather services need to know whether a promising model works at the lead times and in the regions they care about.

The prior constraint

Forecast claims are hard to compare when systems use different variables, periods, and score definitions.

AI’s actual role

Participating machine-learning and hybrid systems made real-time forecasts under common rules for temperature, pressure, and precipitation.

The documented result

The public dataset covers common days 19–25 and 26–32 lead windows and publishes forecast files and regional scoring for JJA 2026.

Why it may matter

Shared evaluation lets weather agencies and researchers inspect where skill appears rather than relying on a single global headline.

Limitations

Benchmark skill does not make a forecast actionable in every region.

A single season is not long-term operational validation.

Warnings and public decisions remain institutional and human work.

Unresolved questions

How stable are rankings across seasons and extremes?

Which regions have too little observation data for confident scoring?

How will participating systems document revisions?