Benefit Ledger · verified · Climate & Environment

Pangu-Weather outperforms IFS on many medium-range scores at far lower cost

Huawei’s Pangu-Weather, published in Nature on 5 July 2023, reported 3D Earth-specific transformer forecasts that beat operational IFS on a majority of 3–7 day upper-air targets while running more than 10,000 times faster.

5 Jul 2023Tier 3 MajorMethodology 0.1

Submit a correction

Current score

+1.73

10 base · Major (tier 3 of 5, 10 pts)

× 0.9000 attribution · Primary causal contribution

× 0.7000 evidence · Peer review or substantial independent validation

× 0.5500 realization · Experimentally validated

× 0.5000 durability · Medium-term

Event-level product before credit split: 1.73

First Nature-level ML model to beat IFS on a broad medium-range scorecard (tier 3). High attribution. Peer review plus later independent MLWP comparisons. Research realization. Quickly surpassed on some scores. Follow-up ingest: added a follow-up source.

What happened

The paper helped trigger the 2023–2025 wave of ML weather models. Forecasts are deterministic and trained on ERA5; operational caveats include precipitation and extreme-event calibration. ECMWF and others later ran independent intercomparisons. Update (2026-09-22): Anthropic Releases Claude Opus 5.5: Fable 5.1-Level Performance at 40% Lower Running Cost Than Opus 5 - MarkTechPost. Anthropic Releases Claude Opus 5.5: Fable 5.1-Level Performance at 40% Lower Running Cost Than Opus 5 MarkTechPost