Docs Model Gallery
The Model Marketplace brings together the time-series foundation models and machine learning methods available in TimechoAI. Filter by model type or search by name, description, and tag to find a model for your data and forecasting scenario.
Browse Models
Select Model Gallery in the left sidebar to open the model list. Each card shows the model name, type, summary, and organization. Click a card or its arrow to open the model details page.
The marketplace currently includes the following model categories:
| Category | Model | Key capability |
|---|---|---|
| Time-series foundation model | Timer-3.5 | MoE foundation model for zero-shot univariate quantile forecasting |
| Time-series foundation model | Timer-3.0 | Generative foundation model for zero-shot probabilistic univariate forecasting |
| Time-series foundation model | Chronos-2 | General-purpose zero-shot forecasting with related series, covariates, and quantile outputs |
| Machine learning method | AutoARIMA | CPU statistical model that automatically selects ARIMA or ARIMAX orders |
| Machine learning method | Holt-Winters | CPU statistical model based on seasonal triple exponential smoothing |
The available models and versions may change as the platform evolves. Refer to the marketplace for the current list.
Filter and Search
- Select All, Time-series foundation models, or Machine learning methods to filter by model type.
- Enter a model name, description, or tag in the search box, then click Search.
- Clear the search and select All to restore the complete model list.
View Model Details
The details page presents the model summary, tags, and basic information so that you can assess whether it fits the task. Depending on the model, the page may include:
- Model type and supported tasks
- Input and forecast lengths
- Parameter scale and release date
- Open-source license and organization
When selecting a model, consider the data length, whether covariates are present, the required forecast length, and the expected output. If you are unsure, choose Auto on the Create Session page and let the system select a forecasting strategy based on the data.
Try Now
Click Try Now in the upper-right corner of a model details page. The platform takes you directly to the Create Session page, where you can add or upload time-series data, review the forecast settings, and run the forecast.
After the redirect, verify that the model and parameters in the session meet your task requirements before submitting the forecast.
Next Steps
- Example Library → — Try forecasting with built-in data or review live application results
- Create Session → — Learn how to provide data, configure parameters, and review results