
WindBorne Models
Built on proprietary balloon observations and continuous AI data assimilation, WindBorne delivers two forecast products: WeatherMesh, our family of AI weather models, and MetaMesh, our multi-model blend.
WeatherMesh
WeatherMesh transforms raw balloon observations into the world's fastest, most accurate weather forecasts. By combining cutting-edge AI models with a unique global network of weather balloons, we deliver insights at a speed and scale that traditional physics-based forecasting cannot match.
Learn more about the latest WeatherMesh model or see our live benchmarks comparing WeatherMesh to ECMWF.
MetaMesh
MetaMesh is WindBorne's multi-model blend, fusing the world's leading AI and NWP forecasts with our proprietary WeatherMesh models. It's designed to consistently outperform the best individual model across the full 15-day horizon.
You can find a list of additional models here.
Compare Models
| Model | wm-6 | wm-6-3km | mm |
|---|---|---|---|
| Code | wm-6 | wm-6-3km | mm |
| Description | Next‑generation global model with 128 members and expanded surface variables. | 2.5 km regional surface forecasts over CONUS. Europe available soon. | Multi‑model blended forecast combining multiple input models with learned bias correction. |
| Type | Continuous Ensemble | Deterministic | Rolling Multi‑model blend |
| Init Frequency | Hourly | Every 15 min | Rolling blend as soon as any input model becomes available. |
| Forecast Horizon | 15 days | ~4 days | 15 days |
| Ensemble Members | 128 members | None (deterministic) | None (deterministic) |
| Gridded Time Step | 3-hourly | Hourly | ✗ |
| Observation Latency | ~1 hour | ~15 min | Varies by input model |
| Spatial Resolution | 0.25° (~25 km) | 2.5 km | Point‑based (no grid) |
| Vertical Levels | Surface + 25 pressure levels | Surface only | Surface only |
| Surface Variables | 40 variables | 8 variables | 9 variables |
| Upper-Level Variables | 5 variables | None | None |
| Point Forecast | ✗ | ✗ | ✓ |
| Interpolated Points | ✓ | ✓ | ✗ |