Pinterest Performance+ creative: feed it asset variations
Pinterest's new Performance+ creative model selects at the asset level. Make distinct hook and image variations in one Sume bulk run and upload the keepers.

Pinterest says its new Performance+ creative model evaluates a broader set of creatives and moves optimization from the ad level to the asset level. For you that means more distinct assets give it more to choose from. Sume can make those variations as one bulk run of up to 100 Format runs, with sume-video-hook for hook swaps.
Pinterest facts are from its June 2026 Cannes post, read 2026-09-30. Sume facts are from Bulk runs and the Format catalog. Pinterest's page does not say how many assets to supply, so this post gives no target number.
What changed in Performance+ creative?
Pinterest describes dynamic creative selection: the model picks the variant most likely to perform for each ad impression. In its testing it reports 7.5% more click volume than the previous single-variant model. That is Pinterest's own number from its own test, not a promise for your account.
What counts as a distinct variation?
Variants that differ in the first seconds, the framing or the message give a selector something to rank; ten near-copies do not. Change one axis at a time so you can read the result: the hook line, the opening shot, or the image crop.
| Axis | Sume tool | Note |
|---|---|---|
| Opening hook | sume-video-hook Format | One run per hook |
| Image crop | aspect_ratio 2:3 or 4:5 | Listed in image docs |
| Message | instruction per item | Same body as a single run |
How do I make them in one request?
Post to /v1/formats/sume/sume-video-hook/bulk-runs with concurrency from 1 to 16 and items of 1 to 100. Each item is a normal run with its own receipt, so one failed item does not cancel the others. Use a fresh Idempotency-Key per batch, then read each child at GET /v1/format-runs/{run_id}. Hook ideas are in hook variations for UGC ads.
What do I do before uploading?
Drop duplicates and anything that misstates the product, then upload the rest through Pinterest's own tools. Keep a note of which hook each file used so you can read results per asset.
Sources
Related posts
More in Use cases
- Turn product photos into a silent video with Timeline static holds
Timeline 1.0 treats a still as a static hold, so product photos become a silent video with fades. Shopify lists MP4 up to 10 minutes; here is the request.
- Pull product stills from a video with the video-frames API
Video frames returns jpeg or png stills from a Sume-hosted clip at set seconds. Google lists 500 x 500 minimum, so extract from the clip before captions.
- Use a frame from your footage as a reference for AI video
Pull a still at a time you pick with video frames (unbilled), then pass it as image_url or reference_image_urls to a Sume video model. Steps and limits.
- Remove background from video with AI: what Sume covers
Sume has no video background-removal endpoint: RMBG 1.0 takes a still image. What you can do instead is a prompt edit with video_to_video, or a crop.
Written by Sume