Ai2 releases HiRO-ACEv2, an AI climate emulator with 3 km resolution

The Allen Institute for AI (Ai2) has released the HiRO-ACEv2 system, which generates global fields of precipitation, 10 m wind, and sea-level pressure at 3 km resolution at a cost roughly 300 times lower than the storm-resolving X-SHiELD model.

Compiled by Sergey KostenchukPublished 2026-10-20Updated 2026-10-20

2026-10-20 Research HuggingFace

🤖 Ai2 releases HiRO-ACEv2 AI climate emulator

ACE computes the atmosphere on a 100 km grid, while the HiROv2 diffusion model "paints in" details down to 3 km: precipitation, 10 m wind, and sea-level pressure. A global run costs roughly 300 times less than the storm-resolving X-SHiELD, and field statistics match it within 3%—including rare extremes and tropical cyclone energy.

🌍 3 km resolution data is needed to study downpours, hurricanes, and atmospheric rivers, but previously such runs were only affordable for NOAA-level centers. Now, multi-year simulations are available to regular research groups.

👤 Weights are open under Apache 2.0 on Hugging Face, code is on GitHub (ai2cm/ace), and there is a demo at ace.allen.ai. Caveats: the model only reproduces X-SHiELD climate in the 66° S — 70° N band and does not replace weather forecasting.

Source 1: https://allenai.org/blog/hiro-acev2 Source 2: https://arxiv.org/abs/2610.17693

Sources