# Export and share

Save an interactive graph snapshot, or export an image for a document or discussion.

## Save a graph file

In **Open Canvas in Browser**, choose **Export graph** to download a `.tensorviz.json` file. It keeps the captured graph and supported saved views and notes. The editor Canvas and the coding-agent plugin's graph offer **Publish** to share through your website account.

Open it with **Open graph file viewer** to explore the snapshot. Recipients do not need to run Python to browse a saved graph.

## Share a view

In the graph-file viewer, arrange the graph and save the view you want someone else to see. You can add a walkthrough or discussions on layers and connections, then download the updated file.

To share through the website, follow [Share to tensorviz.dev](/share-to-website.html).

## Export an image

Use the Canvas export controls for **SVG** or **PNG** when you need a static image.

## Know what you are sharing

A graph file is a snapshot, not your live Python process. It does not execute the model or prove that a later source revision has the same behavior. JAX files preserve inferred shapes and traced-program coverage; they cannot relabel those shapes as runtime measurements. Review the file’s contents before sharing it.

[Explore a graph](/canvas.html) · [Compare code revisions](/graph-diff-review/)

<details>
<summary>Views, discussions, file format, and snapshot comparison</summary>

[Graph file reference](/reference/graph-files.html)

</details>