What We Shipped

Image previews in the review screen are easier to read at a glance.

• No more cropping on resize: previews letterbox-fit when the window changes size instead of getting cut off
• Legible failed-checks overlay: the failed-checks overlay is solid and high-contrast so it stays readable over any image
• Thumbnails in the shot list: each shot in the list now shows a thumbnail

The public tool, use-case, comparison, alternative, and industry pages now connect to each other and present better when shared.

• Pages that link together: compare a tool to others, jump to related alternatives, and find industries that use a pair of tools, all from a single page
• Proper social previews: each page ships its own share image, with tightened titles, descriptions, and absolute canonicals
• Last updated dates on best and compare pages: the /best and /compare pages now show when the page was last refreshed, and the /best pages add a "How we tested" link to the methodology
• A cleaner industry page: the industry page was polished and stripped of its keyword-stuffing chips

The main calls to action on the public pages were sending brand-new visitors to the wrong place, and the free-credit numbers did not always agree with each other.

• Try Free goes to signup: the primary Try Free and Start Free buttons on the best-of and use-case pages were routing new visitors into login, and now go to registration
• One source for the numbers: free-credit counts and model lists come from a single source, so the hero, CTA, and FAQ no longer quote different figures

Choosing the SeedVR upscaler used to run the job on the ESRGAN model instead, so you got the wrong quality and the wrong cost while the job still claimed to be SeedVR. SeedVR upscales now route to the correct endpoint.

• Right model, right result: SeedVR requests now run on SeedVR, not a substitute
• Right cost: you are charged for what actually ran instead of a mismatched price

Your saved workflows no longer need a manual run. You can now set them to fire on their own when something happens in your workspace.

• Four trigger types: start a workflow on new uploads, on product changes, on a schedule, or from an authenticated webhook
• Draft and preview before you commit: build a trigger as a draft, then dry-run it against recent uploads or product changes to see what it would have matched
• Enable, disable, and duplicate inline: flip a trigger on or off from the list, or duplicate one as a disabled draft without copying its credit caps or webhook secrets
• A full attempt history: every trigger keeps a record of recent runs, including the product, changed fields, and source context behind each attempt

The public site grew several new sections to help you size up Studio before signing in.

• Interactive pricing calculator: get a recommended plan from your asset, vendor, and AI-usage estimates
• Free tools hub: a dedicated place to find the free tools in one spot
• Customer stories: a customers section with case studies showing how teams use Studio
• Contact by intent: a contact page split into demo, sales, support, press, and partnership
• A fuller pricing FAQ: the pricing FAQ went from 4 to 12 entries covering credits, plan changes, refunds, seats, and taxes

The GPT Image model behind AI generation and editing moved from GPT Image 1.5 to GPT Image 2, with stronger text rendering and prompt adherence.

• Newer model everywhere it is used: generate, image-to-image, and image-translation tools all run on GPT Image 2 now
• Pick GPT Image, get the new one: choosing the GPT Image model gives you the upgraded version automatically
• Old ids retired: existing flows that referenced the old model now resolve to GPT Image 2, so nothing breaks

Studio now has a dedicated reviewer team role, and reviewers can claim and hand off work in the approval queue.

• A role of its own: reviewers get their own invite flow, separate from suppliers and full members
• Only the tools they need: the review tab and queue are hidden from teammates who lack approval access
• Claim and assign: reviewers can claim submissions and assign or unassign them, with reviewer badges in list rows and the detail panel
• Focus on your work: the queue defaults to showing the submissions assigned to you, and the sidebar shows how many are waiting on you

The library list view got a real tune-up. Moving columns around, resizing them, navigating with the keyboard, and working through big folders should all feel smoother now, especially when the grid has a lot on screen.

• Column drag and resize are steadier: less jumping, fewer weird snaps, and much better behavior after auto-fit or reset
• Scrolling does less work: the grid renders less offscreen content, trims hover chrome during fast scroll, and avoids mounting row actions everywhere
• Selection and row states stay lined up: pinned columns, group rows, and selection borders stay in sync while you move around
• Name links are sane again: in list mode, folder and asset names are clickable only on the text, not across the empty space to the right

Supplier and reviewer assignment management now feels like one system instead of a pile of similar-looking screens.

• The same assignment flow everywhere: suppliers and reviewers now share the same add and edit patterns, the same assignment cards, and the same searchable pickers
• Less scope clutter: assignment options are trimmed down to All products, Product type or Category fallback, Brand, plus a separate Specific products mode when you need to target exact SKUs
• Better product search for multilingual catalogs: product lookups now normalize accents and rank exact phrase and all-token matches ahead of fuzzy fallback, which fixes searches that used to fail once you typed the full query