GenCast: Diffusion-based ensemble forecasting for medium-range weather
Summary
GenCast is a probabilistic machine-learning weather model trained on decades of reanalysis data. It generates an ensemble of stochastic 15-day global forecasts, at 12-hour steps and 0.25 degree latitude-longitude resolution, for over 80 surface and atmospheric variables, in 8 minutes. The authors report greater skill and speed than ECMWF's operational ensemble forecast, ENS, including greater skill than ENS on 97.4% of 1320 evaluated targets, and better predictions of extreme weather, tropical cyclones, and wind power production. [S1]
Claims
- id: C1 claim: 'GenCast generates 15-day global ensemble forecasts at 12-hour steps and 0.25 degree resolution, in 8 minutes.' source: S1 evidence: "stochastic 15-day global forecasts, at 12-hour steps and 0.25 degree latitude-longitude resolution"
- id: C2 claim: 'The forecasts cover over 80 surface and atmospheric variables.' source: S1 evidence: "for over 80 surface and atmospheric variables, in 8 minutes"
- id: C3 claim: 'The authors report greater skill than ENS on 97.4% of 1320 evaluated targets.' source: S1 evidence: "greater skill than ENS on 97.4% of 1320 targets we evaluated"
Verification
- Code available: No official GenCast model or training-code URL was named on the abs page or HTML.
- Reproduced: not reproduced (2026-11-12). Code not run.
- Related work: Live note 2212-12794-graphcast.
- Limits: Comparisons against ECMWF ENS are author-reported (v2, 1 May 2024). No official code URL was found in the abs page or HTML.
- License: CC BY 4.0 on the abs page (2026-11-12; icon by-4.0.png; http://creativecommons.org/licenses/by/4.0/). Submitted 25 Dec 2023 (v1); last revised 1 May 2024 (v2). Sources: [S1] https://arxiv.org/abs/2312.15796, v2 (1 May 2024). checked_on: 2026-11-12.
When to apply
Use when comparing learned ensemble weather forecasts with GraphCast-style deterministic models or operational ENS.
Facts
- id: F1 claim: 'GenCast is a diffusion-based ensemble weather model that produces 15-day global forecasts at 0.25 degree resolution in 8 minutes.' confidence: medium date: 2026-11-12 sources: [S1] license: CC BY 4.0
- id: F2 claim: 'The forecasts include over 80 surface and atmospheric variables at 12-hour steps.' confidence: medium date: 2026-11-12 sources: [S1] license: CC BY 4.0
- id: F3 claim: 'The authors report greater skill than ECMWF ENS on 97.4% of 1320 evaluated targets.' confidence: medium date: 2026-11-12 sources: [S1] license: CC BY 4.0
Sources
- [S1] arXiv 2312.15796, https://arxiv.org/abs/2312.15796. License: CC BY 4.0. v2 (1 May 2024). Accessed 2026-11-12.
License: this note is CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Summary by iinventedlight (AI-assisted). Paper is CC BY 4.0.
License: CC-BY-4.0 · Summary by iinventedlight (AI-assisted). Source: Ilan Price, Alvaro Sanchez-Gonzalez, Ferran Alet, and 9 coauthors (listed on the arXiv abs page), GenCast: Diffusion-based ensemble forecasting for medium-range weather (CC BY 4.0).
Cite this
'iinventedlight (AI-assisted)'. 2026. "GenCast: Diffusion-based ensemble forecasting for medium-range weather." iinventedlight Library, iil:note:83ctfzhr, version of 2026-11-12. https://iinventedlight.com/library/notes/2312-15796-gencast/. License: CC-BY-4.0.
@misc{iil_note_83ctfzhr,
author = {{'iinventedlight (AI-assisted)'}},
title = {{GenCast: Diffusion-based ensemble forecasting for medium-range weather}},
howpublished = {iinventedlight Library},
year = {2026},
date = {2026-11-12},
url = {https://iinventedlight.com/library/notes/2312-15796-gencast/},
note = {Permanent ID iil:note:83ctfzhr. License: CC-BY-4.0.},
}
Spotted a mistake? Use the MCP feedback tool or email hello@iinventedlight.com.
For language models and agents: start at /llms.txt | /agents.md | /.well-known/agent-library.json. This page in Markdown: /library/notes/2312-15796-gencast.md.