What We Shipped

Full digital asset management dashboard with a new sidebar layout.

• Activity sidebar tab with full-screen view in the dashboard
• Flows tab for managing workflow orchestrator runs
• Right-click context menu on gallery items (rename, move, delete, process)
• Analytics section with usage dashboard

A new Batch Generate tool creates many images at once instead of one prompt at a time.

• Many prompts, one run: feed a list of different prompts and get an image for each
• One prompt, many variations: keep a single prompt and generate multiple variations of it
• Single or Batch: a switcher moves you between generating one image and running a whole batch

New tool for optimizing GLB/glTF 3D models for the web. Reduce file sizes by compressing textures and geometry while preserving visual quality.

• Single and batch mode — optimize one model or drop a folder of GLBs
• Server-side processing via Inngest for large files
• Vercel Blob upload with optional Sirv CDN promotion
• Inline 3D preview so you can inspect results before downloading

Official WordPress plugin for Sirv AI Studio. Connect your WordPress site and run AI tools on your media library without leaving the WP admin.

• OAuth authentication — secure connection without storing API keys in WordPress
• Zapier batch endpoints for alt text generation and image creation
• Background removal, upscale, and alt text available from the WP media library
• Docker dev setup included for local testing

The prompt expander now analyzes your uploaded image to write context-aware expanded prompts. Moved to the top-right corner and available on all tool pages.

Workflow pipelines are no longer limited to linear chains. DAG (directed acyclic graph) mode lets you branch, merge, and create complex multi-path workflows.

• Free-form connections — drag edges between any compatible nodes
• Auto-connect on add — new steps auto-wire to the selected node when compatible
• Session restore — DAG layouts and custom connections persist across page reloads
• Linear mode preserved — simple chains still work the same way, DAG is opt-in via the canvas toolbar

Two model updates give you more reach on resolution and finer control over re-shooting a product from a different viewpoint.

• SeedVR upscaling: the Upscale tool adds a new SeedVR model with up to 10x scaling and target-resolution presets from 720p to 2160p. It is the most affordable upscaler, at 1 credit.
• Better camera-angle control: the Multiple Angles model in the image editor now uses precise rotation (0 to 360 degrees around the object), elevation (low-angle to bird's-eye), and zoom (wide to close-up).
• Re-frame without a re-shoot: change the viewpoint of an existing product image instead of setting up a new photo.

New autofix node for the workflow orchestrator. It runs AI review on images and automatically fixes issues it finds, then sorts results into fixed vs. unfixable categories.

• Branching logic so fixed images continue down the pipeline
• Categorized outputs to separate fixed and unfixable images
• Review results pass through to downstream nodes

A new Virtual Try-On Video tool creates a video of a model wearing your garment, so you can show how a piece moves rather than just how it sits in a flat photo.

• Multiple garments per video: dress a model in more than one item in a single clip
• Runs in the background: generation happens as a background job, so you can keep working while it renders
• Built for fashion catalogs: turn a static garment shot into a moving try-on without a real shoot

A new Marketplace Optimizer checks a product image against marketplace requirements and tells you whether it will pass before you list it.

• Compliance scorecard: see at a glance how an image measures up to the rules for marketplaces like Amazon and Shopify
• Suggested fixes: get concrete optimisations for what is holding the image back
• Catch problems early: review an image against the rules before upload instead of after a rejection