Latest version: 26.23 (released August 31, 2026). App Store release notes.




The mobile app is only available for Multi-tenant Cloud accounts. It isn’t available for Dedicated Cloud or Self-managed deployments.
Download the app
Install the app from the App Store to get started.Download on the App Store
Features
The following features are available in the mobile app.- Explore projects: Browse and search across your W&B projects.
- Full project history: Metric charts on project panels include data from all runs in the project, not only the latest run.
- Star projects: See your most important projects at a glance. Tap the star icon next to a project to star it. To filter the list to only starred projects, tap the Starred tab at the top of the list.
- Track experiments: View run status, metrics, and line plots for your experiments in real time.
- Live updates: Charts refresh as new data arrives. The Runs tab and project run lists update automatically when new runs start.
- Run overview tab: On each run, open the Overview tab for a dedicated summary of run details alongside metrics and logs. View the full run configuration and summary, including nested objects and arrays. Long config values wrap to multiple lines. Long-press a configuration value to copy it.
- View system metrics: On each run, open the Metrics tab to see how your hardware behaved during training. W&B automatically logs metrics such as GPU utilization, CPU usage, memory, disk I/O, and network traffic. System metrics appear in a collapsible System section alongside your logged metrics. Each metric displays as a line chart with the same tooltip and search behavior as other metrics on the tab.
- Mobile-optimized panel grouping: Panels are automatically grouped by name into collapsible sections. You can collapse the ungrouped metrics section to focus on named sections. Sections are a single level (not nested), and grouping follows the same rules as workspace panels in the W&B web app, so the layout stays consistent when you move between desktop and your phone.
- Star panels: See your most important panels at a glance. When viewing a run or a project, tap the star icon at the top of a panel to star it. To filter the list to only starred panels, tap the Starred tab at the top of the list.
- Star metrics: When viewing an important metric in detail view, tap the star icon to star it. Run cards on the Runs page prioritize your starred metrics when choosing which metric to display.
- Search panels: When viewing a run or a project, use the panel search field at the bottom of the screen to filter which runs appear in each chart. You can search with RE2 syntax to match patterns in run names.
- Chart tooltips: On line charts, tooltips show metric values with up to four decimal places so you can read small changes accurately.
- Zoom charts: Pinch vertically or horizontally on a line chart to zoom in on either axis, or double-tap to step through zoom levels. The Y-axis rescales automatically to fit the visible portion of the curve, so small variations late in a run stay readable.
- View image metrics: Images you log to a run appear on the Metrics tab as panels you can interact with. Tap to view a single image, then pinch to zoom or drag to pan. See View image metrics.
- Change line plot settings: Switch a line plot to a logarithmic Y-axis and apply smoothing, for a single plot or for every plot in a run or project. See Change line plot settings.
- Filter metrics: Use the metric filter controls to narrow which metrics appear, so you can focus on the ones that matter.
- Stop runs: When viewing an in-progress run, tap the action () menu, then tap Stop run.
- View console logs: Read logs from active or completed runs, or download them once a run finishes. Colors from ANSI escape codes in your script output render faithfully and adapt to light and dark mode. The app polls active runs for new log lines every 5 seconds. Polling stops once a run finishes or crashes, which saves network and battery use. Even for runs with hundreds of thousands of log lines, the most recent output appears immediately, and older lines load in the background. Scrolling and log search stay responsive even for large outputs.
- Switch teams: Use the team picker to change teams.
- Stay informed: Get updates on your experiments without opening your laptop using notifications. Remove alerts you no longer need from the Notifications tab by swiping them away.
- Chat with ARIA: Use ARIA in the mobile app to ask questions, analyze runs, and take action without leaving your phone.
View image metrics
Images you log to a run withwandb.Image appear on the run’s Metrics tab, where each image metric has its own panel alongside your line plots. The following actions are available for image metrics.
- Scrub through steps: Drag the step slider below the images to move through the steps the metric logged. When the metric logged only one step, the slider doesn’t appear.
- Jump to a step: Tap the step value next to the slider, enter a step, then tap Go to step. The sheet shows the range of steps the metric logged, and your entry snaps to the nearest logged step.
- Compare images within a step: When a step logs more than one image, the panel arranges them in a grid that you swipe horizontally to page through. Dots below the grid show which page you’re viewing.
- View an image full screen: Tap the panel to open the detail view, then tap an image. Pinch to zoom and drag to pan, and swipe to move to the next or previous image in the step.
- Star image metrics: Tap the star icon on the panel to star an image metric, the same as other panels.
Change line plot settings
Line plot settings control how the app draws a line plot. Open the settings in one of the following ways:- On a run’s Metrics tab or a project’s Panels tab, tap the line plot settings button next to the metric filter. Changes apply to every line plot in that run or project.
- On a single chart’s detail screen, tap the line plot settings button in the toolbar. Changes apply only to that plot.
- Log scale (Y): Draw the Y-axis logarithmically to compare a metric across orders of magnitude. A logarithmic axis can’t plot zero or negative values. If a plot has no positive values to draw, the app replaces the chart with the message “No positive values to plot on a log scale”.
- Smoothing: Apply time-weighted exponential moving average (TWEMA) smoothing, from
0(off) to0.99. This algorithm is also the default in the W&B web app.
Chat with ARIA
ARIA (AI Research & Iteration Agent) is your personalized research assistant in W&B. Use ARIA for tasks such as analyzing and running experiments, explaining results, identifying patterns across runs, recommending next steps, and building visualizations and reports. To chat with ARIA, tap- Project context: ARIA automatically detects which project you’re viewing in the app and uses it as context, so answers come with the right project already in mind. You can also reference projects and runs directly in your chat messages.
- Rich replies: Replies render with rich Markdown formatting, including tables, headings, lists, and inline code. When ARIA asks questions, they render as an interactive view directly in the conversation.
- Visible actions: Shell steps show when ARIA takes an action, such as creating a panel, filtering runs, creating and running Python scripts, and more.
- Chat history: Long-press a chat in your chat history to preview, rename, or delete it. Chat history is kept in sync so you can move seamlessly between the web app and mobile app.

Set up notifications
Configure notifications to receive updates about your experiments on your mobile device.
Metric threshold alerts
Use metric threshold alerts to get notified when a metric crosses a value you specify.- Navigate to a run.
- Tap the graph to enter full screen view.
- Tap the bell icon in the top right.
- Set notifications that trigger when future runs cross the specified metric threshold.
Run failure alerts
Use run failure alerts to get notified when runs in a project fail.- Navigate to a project.
- Tap the action () menu in the top right.
- Select Run failed alert to receive notifications for future run failures in that project.