# Timestep-Conditioned Transformers for Global Weather Forecasting

Preprint · Sam Levang, Fran Bartolic, Ty Dickinson, Chase Dwelle, Paulius Rauba, Viktor Cikojevic

- Archive id: arxiv:2608.06241
- Published: 2026-08-06
- arXiv: https://arxiv.org/abs/2608.06241
- Citations: 0
- Source URL: https://arxiv.org/abs/2608.06241
- PDF: https://arxiv.org/pdf/2608.06241
- Provenance: arxiv

## Abstract

Existing machine-learning weather forecasting models rely on predetermined and fixed autoregressive timesteps. The choice of model timestep involves a fundamental trade-off: shorter timesteps (e.g. 1 to 6 hours) finely resolve atmospheric dynamics within the diurnal cycle but increase error accumulation for a given forecast horizon, while longer timesteps (e.g. 24 hours) reduce error accumulation but limit the usability of short-range forecasts where sub-daily predictability is high. In this work, we present GEM-3, a probabilistic global weather model that addresses this trade-off through explicit multi-timestep inference. With a single set of trained weights, the model timestep can be configured at inference time to balance predictability and usability across a broad forecast horizon. Additionally, we find that mixed-timestep training consistently improves rollout stability relative to timestep-specialist models. Under the hood, GEM-3 is a lightweight neighborhood-attention transformer with ~134M parameters on an equirectangular grid with a number of architectural advancements beyond its predecessor GEM-2. The result is a practical forecasting system that couples near-SOTA medium-range probabilistic skill, stable extended-range rollouts, efficient training and inference, and decision-relevant diagnostics.

_Abstract reproduced under CC0-1.0 (arXiv descriptive metadata)._

Topics: [AI Weather & Climate Modeling](https://space-kitty.com/topics/ai-weather-climate)

---
Space Kitty Archive · https://space-kitty.com · API: https://space-kitty.com/api/v1 · MCP: https://space-kitty.com/mcp · Premium: $0.20/request via x402 (Base Sepolia (testnet)). Metadata: arXiv (CC0), OpenAlex (CC0), Europe PMC (CC BY/CC0 works), ClinicalTrials.gov.