Pixelcut teardown: 13.64M visits a month for the AI product-photo editor that sells credits to sellers
Monthly visits
13.64M
Semrush, October 2026. Third-party estimate, not official MAU/DAU.
Pixelcut draws an estimated 13.64M visits a month (Semrush, October 2026). That makes it the forty-sixth largest site in our database of 195, just below Mailchimp at 14.24M and just above Wordcounter.net at 13.24M, and the fourth largest AI photo editing site we track, behind Picsart at 25.67M, PhotoRoom at 24.2M and Pixlr at 16.21M. It is the thirteenth image or PDF tool we have torn down. The product is a free AI photo and video platform built for sellers: background removal, AI backgrounds and batch editing, plus a Creative Agent that plans and executes jobs with third-party models like Nano Banana Pro and Sora 2. The paid tiers sell AI credits — 600 a month on Pro, 3,600 on Business — and the operator is Pixelcut Inc., a Y Combinator Summer 2020 company based in San Francisco.
| Site | Pixelcut, pixelcut.ai |
|---|---|
| What it is | Free AI photo and video platform for sellers: background removal, AI backgrounds, batch editing, AI image and video generation, and a Creative Agent that plans and executes tasks |
| Who it is for | E-commerce sellers, brands and marketing teams who need product images and videos at volume without a studio |
| Category | AI photo editing |
| Monthly visits | 13.64M, Semrush, October 2026. Third-party estimate, not official MAU/DAU |
| Pricing | Free, plus two paid tiers. Pro $10 a month billed monthly or $8 a month billed annually; Business $30 a month or $24 a month billed annually. Annual billing saves 20%. Source: Pixelcut pricing page, read 16 November 2026 |
| Pricing model | Freemium with AI credits. Pro buys 600 AI credits a month, Business 3,600. Credits are spent on third-party AI models, and the pricing page publishes a per-model table of how many images or videos each plan's credits buy |
| Country breakdown | India 20.11% (2.74M); United States 16.9% (2.31M); Indonesia 11.55% (1.57M). Semrush, October 2026 |
| Operator | Pixelcut Inc., San Francisco. Y Combinator Summer 2020 company; co-founders Dominique Yahyavi and Prasanth Veerina. Sources: master-table record (round 5); company press release, March 2026; third-party company profiles, read 16 November 2026 |
| Acquisition channels | Not publicly disclosed (see the acquisition section) |
TL;DR
- Pixelcut draws an estimated 13.64M visits a month (Semrush, October 2026), the forty-sixth largest site in our database and the fourth largest AI photo editing site, behind Picsart at 25.67M, PhotoRoom at 24.2M and Pixlr at 16.21M. India is first at 20.11% (2.74M), the United States second at 16.9% (2.31M) and Indonesia third at 11.55% (1.57M). The top three together are 48.56%. India alone sends 2.74M visits, more than the whole traffic of CombinePDF at 2.5M.
- The meter is AI credits, spent on other companies' models. The free tier gives limited background removal and limited upscale, with free export without a watermark. Pro costs $10 a month billed monthly or $8 a month billed annually and buys 600 AI credits a month; Business costs $30 or $24 and buys 3,600. The pricing page publishes a per-model conversion table, and the same credits buy wildly different amounts depending on the model: on Business, Nano Banana Pro buys up to 6,000 images, Flux 2 Pro up to 15,000, ChatGPT Image 2 up to 1,111, and Kling 4.0 up to 1,028 videos.
- It sells to sellers, not to everyone. The homepage calls itself "The AI Video and Image Platform" and opens with "Join 70 million sellers making images and videos with AI", then organises the product around product showcases, AI UGC ads and brand-consistent personas, with a brand wall of Adidas, Apple, Nike, Sephora and Toyota. The risk in the story is that it rents its models, the credit conversion is hard for a newcomer to compare, and third-party review trackers report a 1.7 out of 5 Trustpilot score and a BBB F rating — aggregator reports, not our verification — while the company publishes no revenue, no funding and no acquisition mix.
What problem does Pixelcut solve, and for whom?
The problem is the product photo. A seller with a hundred listings needs a hundred clean, consistent images, and a studio shoot for each one is slow and expensive. Pixelcut's homepage states the audience in the first line after the headline: "Join 70 million sellers making images and videos with AI." The 70 million figure is the site's own claim, quoted with that label attached, and it is doing the same work as the brand wall underneath it — Adidas, Airbnb, Apple, Chanel, Google, Gucci, H&M, Nike, Sephora, Starbucks and Toyota — which is to tell a small seller that big brands use the same tool. The master-table record puts the audience the same way: e-commerce sellers and marketing teams who need product images at volume.
The positioning is narrower than its neighbours'. Picsart sells to every creator and Pixlr sells to anyone with a photo; Pixelcut sells to the person with a product to list. That narrowness shows up in the traffic shape: India first at 20.11%, the United States second at 16.9% and Indonesia third at 11.55%, with the top three together at 48.56%. It is the only AI photo editing teardown where the United States is not first, and the India-first pattern fits a seller base that is heavily weighted toward marketplaces and small exporters rather than US creators.
The pitch is that the whole listing workflow lives in one place. The homepage groups the product into three workflows — Product Showcase, AI UGC Ads and Personas — and then into everyday edits and on-brand output, so the seller's job is framed as a pipeline from one product photo to a finished ad, not as a single edit. The free tier is the front door to that pipeline: limited background removal and limited upscale, but free export without a watermark, which is the one free-tier detail that separates Pixelcut from most of its category, where watermarked export is the usual paywall.
The product and its first screen
The first screen sells an agent, not a tool. The headline is "The AI Video and Image Platform", and the section under it is "Smarter than your average tool": "Pixelcut understands your prompts, plans and executes tasks on your behalf using the best creative tools available like Nano Banana Pro and Sora 2. No special prompting required." That is the 2026 version of the product: the user describes the job and the agent picks the model. The three workflows under it are the jobs a seller actually repeats. Product Showcase: "Bring your products to life with studio-grade imagery, generated with just a prompt". AI UGC Ads: "Create talking videos to promote your brand in seconds". Personas: "Create custom personas that can be consistent across your content".
The everyday edits are the original product, still on the page. Background Removal is described as ultra-high quality and proprietary, for image and video. Upscale goes up to 16x resolution, "choose from a range of models including Topaz". Expand reframes images with a proprietary tool that extends beyond the edges of the frame. Retouching is "precise AI retouch" that can erase or edit with generative fill. The on-brand section is the retention layer: AI Actors, AI Product Photos ("Generate unlimited product images with perfect character consistency across every shot") and On-Brand Videos, where "your character stays consistent from frame to frame".
The social proof is the site's own display, quoted with that label attached. "Millions of images created every day" sits above a strip of example images. Three testimonials carry the page: "My favorite app — Great app, easy to use. Helps our small business save time and money to create different marketing materials professionally. Thank you." "Best background remover — I need a good background remover for my work and I've been through them all it feels like. This one is the best by far. It's so easy to use and the results always look amazing. Thank you Pixelcut!" "Love it — It's amazingly easy and gives the perfect results for my bags and accessories. It looks like it was done by a professional photographer." The closing line repeats the scale claim: "Over 70 million creators, brands and businesses have trusted us to be their creative workspace". The pricing page carries a smaller, different number — "Trusted by over 200,000 creators, brands and small businesses" — and this teardown prints both, with their pages attached, rather than reconciling them.
Pricing: the meter is AI credits, spent on rented models
There are two paid tiers above the free tier, and the difference between them is mostly a number of credits. Free costs $0 and gives limited background removal, limited upscale and free export without a watermark. Pro costs $10 a month billed monthly or $8 a month billed annually, and buys 600 AI credits a month, access to all AI models, unlimited background removal, unlimited upscale, a 3-person team, 1,000 batch exports a month and a commercial licence. It is marked "Most Popular". Business costs $30 a month or $24 a month billed annually, and buys 3,600 AI credits a month, a 10-person team, 2,000 batch exports a month and a commercial licence. Annual billing saves 20% on both tiers, and the pricing page states that saving directly: "Save 20% with a yearly plan".
The comparison table on the pricing page is where the meter gets specific. Generative Fill, Expand Image and Pro Eraser are limited on Free, 300 a day on Pro and up to 1,200 a day on Business. AI Backgrounds is 300 a day on Pro and up to 1,200 a day on Business. The credits row reads 600 a month on Pro and "up to 180,000" a month on Business — that 180,000 is not a second allowance, it is the cheapest end of the per-model conversion table, and it is the clearest sign that a Pixelcut credit is not a fixed unit.
The conversion table is the most honest page in the category and the hardest to read. The pricing page explains that credits are spent "whenever you create or edit images with certain advanced AI models — particularly those provided by third-party partners", then lists what each plan's credits buy, model by model. On Business: Nano Banana 2.1 up to 12,000 images, Nano Banana Pro up to 6,000, Nano Banana Lite up to 18,000, Flux 2 Pro up to 15,000, Flux 2 Klein 4B up to 180,000, ChatGPT Image 2 up to 1,111, Ideogram v3 up to 10,000, Recraft v4 up to 18,000. Video is priced by the clip: Kling 4.0 up to 1,028 videos, Veo3.1 up to 450, and the page notes that "credit usage for video models are based on 5 seconds of generated video". The same 3,600 credits therefore buy 450 Veo clips or 180,000 Flux images, a 400-to-1 range that the buyer has to work out from the table.
The comparison inside the vertical is short. Picsart charges Pro at $15 a month billed monthly or $10.50 a month billed annually, with 500 credits a month. PhotoRoom charges Pro at $12.99 a month or $7.50 a month billed annually, with 8,000 AI credits and 1,000 exports. Pixlr's Premium is $9.99 or $7.99 a month, with 1,000 credits. Pixelcut's Pro at $10 or $8 a month, with 600 credits, sits in the middle on price and near the bottom on credits — but its free tier is the only one of the four that exports without a watermark, and its Business tier at $30 or $24 is the cheapest team step of the four. The pricing page as read on 16 November 2026 lists no free-trial length; the free tier is the trial, and this teardown prints no trial number.
How does Pixelcut get its users?
Not publicly disclosed. Pixelcut publishes no acquisition mix, and our database holds no verified channel data for it, so this section is short on purpose. Anyone who tells you its traffic is exactly X percent organic search is guessing, however plausible the guess sounds.
Five things are observable without inventing numbers. First, the seller wedge is a search wedge. Every tool on the homepage is a job a seller types into a search box — remove background, upscale, expand, retouch, AI product photo — and each one is its own landing surface. Second, the free tier is the distribution, and it is unusually generous. Free export without a watermark removes the usual reason to pay, which only makes sense if the free tier's job is to spread the tool rather than to convert. Third, the brand wall is the trust shortcut. Eleven household names on the homepage do the work of a case study without a single number attached. Fourth, the model names are distribution. Listing Nano Banana Pro and Sora 2 on the homepage puts Pixelcut on the same search pages as the models people are looking for, the same move Pixlr makes with Flux and Kling. Fifth, the geography is the tell. India first, Indonesia third, and the United States second is the traffic shape of a product that spreads through marketplaces and small sellers, not through US creator marketing.
From visit to payment: the credits meter does the selling
The funnel is the credits meter, with batch work as the second push. A free user arrives, removes a background or upscales an image, and hits the edge of the free tier: unlimited background removal, unlimited upscale, all AI models and 600 credits a month sit behind Pro, and batch exports — 1,000 a month on Pro, 2,000 on Business — sit behind the same paywall. The cheapest step is Pro at $10 a month billed monthly or $8 billed annually, and the annual saving is stated on the page, which is an anchor designed to make the first payment feel like the careful choice rather than the expensive one. The conversion table does quiet selling of its own: by showing that the same credits buy 6,000 Nano Banana Pro images or 1,028 Kling videos, it teaches the buyer to think in credits before paying, and the meter is visible in the same units the user already spent.
Two details in the pricing copy are doing quiet work. The first is the team seat. Pro includes a 3-person team and Business a 10-person team, so the upgrade is also a collaboration decision, and a seller who brings two colleagues has already chosen the paid tier. The second is the commercial licence, which is listed on both paid tiers and not on Free — a seller who wants to use the images commercially has a reason to pay that has nothing to do with credits. What we cannot see from outside is the rate at which free users cross to paid, or how many free users hit the background-removal limit first, because the pricing page lists no free-tier credit number. This teardown does not print one.
Retention: what keeps people coming back?
Not publicly disclosed. Pixelcut publishes no churn, renewal or cohort figures in the sources we checked, so this section states what the product does to support return visits and stops there.
Three features of the product support return visits. The first is the recurring job. A seller lists new products every week, and each listing needs the same background removal, the same AI background and the same batch export, so the next job is already scheduled by the business, not by the tool. The second is the monthly credit refill. A subscription that refills every month is a reason to open the editor before the credits expire, and the reset is a calendar event the product does not have to advertise. The third is brand consistency. Personas and AI Actors keep the same character across every shot, and once a seller's catalogue has a consistent look, switching tools means rebuilding that look. The counterweight is that there is no lock-in. The product exports standard image formats, so a user can leave with their work, and a rival that rents the same models can offer the same output. The only defence is being the editor the seller already knows.
Why Pixelcut is still standing
Our read, stated as a position you can argue with: Pixelcut survives because it picked the narrowest wedge in the category and then rented the best models instead of training any. It began as a background removal tool — the company's own March 2026 press release says "what began five years ago as a background removal tool" — and it is a Y Combinator Summer 2020 company, Pixelcut Inc. of San Francisco, co-founded by Dominique Yahyavi and Prasanth Veerina. The founding year is not printed as a single number in this teardown because the sources disagree: one company database says 2013, the press-release arithmetic and a review profile point to around 2021, and the YC batch is Summer 2020. The press release also records a rebrand: in March 2026 the company renamed itself Pixa and moved from pixelcut.ai to pixa.com, and a third-party review tracker records that it reverted to Pixelcut months later, with no ownership change. The homepage today is pixelcut.ai, and the product is called Pixelcut.
The second half of the argument is the model list. Pixelcut aggregates Nano Banana, Flux, ChatGPT Image, Ideogram, Recraft, Kling, Veo and Sora, which means it never pays to train a model and never has to be the best model of the month. It only has to be the shelf the best model sits on, and the credits meter turns someone else's inference bill into a subscription. The seller wedge is the distribution for that shelf: a product photo is a recurring, measurable job, and the free tier exports without a watermark, which is the one free-tier detail no neighbour matches. The traffic figure backs the shape: an estimated 13.64M visits in a single month makes Pixelcut the forty-sixth largest site in our database, and India alone sends 2.74M visits, more than the whole traffic of CombinePDF. The caveat sits next to the number, as it does on every page of this site: the figure is a single third-party estimate for the month quoted, and Pixelcut publishes no official MAU or DAU.
The last part of the argument is the price of the first payment. A $10 a month step, $8 billed annually, is low enough to be an impulse for a seller who just made a listing, and the credit refill gives the buyer a reason to use it. The risk in that shape is the same as the strength. A product that rents its models can be undercut by anyone who rents the same models, and the only defence is being the editor sellers already know — which takes years to build, a rebrand to test, and a faster free rival to lose.
What you can copy, and what you cannot
Copyable this week
- Pick a wedge narrower than the category. "Product photos for sellers" is a job with a buyer attached, and every page of the product is organised around it, not around editing in general.
- Meter the expensive part, not the seat. AI credits map directly to inference cost, and the pricing page publishes the per-model conversion table, so the buyer can see what the allowance buys before paying.
- Rent the models, do not train them. Listing the best outside models on the homepage keeps the product current without a training bill, and the credits meter passes the cost through honestly.
- Make the free tier genuinely free. Free export without a watermark is the one free-tier detail in this category that costs the company money and buys it distribution.
- Sell the team, not just the credits. A 3-person team on Pro and a 10-person team on Business turn the upgrade into a collaboration decision.
Not copyable
- Five years of being the background remover. A product that began as one tool and grew into a platform has years of links, bookmarks and typed-in visits, and a new tool cannot buy that history.
- The Y Combinator batch and the brand wall. A Summer 2020 batch and eleven household names on the homepage are trust a new product cannot print for itself.
- The seller traffic shape. India first, Indonesia third, the United States second is a distribution pattern that follows marketplaces and small sellers, and it compounds with every listing.
Risks and gaps in this teardown
- The traffic figure comes from a single source for the month quoted, Semrush, and is a third-party estimate. Pixelcut publishes no official MAU or DAU. The October 2026 figure is the same month used for most of the database.
- The founding year is not printed as a single number because the sources disagree: one company database says 2013, the March 2026 press-release arithmetic ("began five years ago") and a review profile point to around 2021, and the Y Combinator batch is Summer 2020. The operator (Pixelcut Inc., San Francisco, YC S20) and the co-founders (Dominique Yahyavi and Prasanth Veerina) are from the master-table record and the March 2026 press release.
- The March 2026 rebrand to Pixa and the move to pixa.com are from the company's own press release of 3 March 2026. The reversion to Pixelcut months later, with no ownership change, is from a third-party review tracker (July–August 2026), not from the company, and is flagged as such.
- Third-party review trackers report a 1.7 out of 5 Trustpilot score ("Bad") and a BBB F rating for six complaints left unanswered, with complaints about countdown-timer pressure and charges after an advertised trial. These are aggregator reports, not our verification, and this teardown does not independently verify them. The pricing page as read on 16 November 2026 lists no trial length.
- The credit conversion is published but hard to compare: on Business, the same 3,600 credits buy up to 180,000 images with the cheapest model (Flux 2 Klein 4B) and up to 450 videos with Veo3.1. The per-model table is from the pricing page, read 16 November 2026. The "up to 180,000" in the credits row is the cheapest-model maximum, not a separate allowance.
- "Join 70 million sellers", "Over 70 million creators, brands and businesses" and "Millions of images created every day" are the site's own claims from its homepage, read 16 November 2026, quoted with that label attached. The pricing page's "Trusted by over 200,000 creators, brands and small businesses" is a different number on a different page, and this teardown prints both rather than reconciling them.
- The pricing page lists no free-tier AI credit allowance, so this teardown prints none. The free tier is described on the pricing page as limited background removal, limited upscale and free export without a watermark.
- Acquisition mix, conversion rates and retention are not publicly disclosed. Sections above say so rather than estimating them.
- Country breakdown comes from Semrush for October 2026 and covers the top three countries only. The absolute figures (2.74M, 2.31M, 1.57M) are the master-table record values. The remaining share is not broken out in the sources we checked.
- Pixelcut is the thirteenth image or PDF tool teardown. The similar-sites table below leads with the closest pattern match: Picsart (25.67M), PhotoRoom (24.2M) and Pixlr (16.21M) are the three larger AI photo editing sites, and Cutout.pro (8.74M) is the next one down, all four selling AI credits or subscriptions next to a free tier.
Similar sites in the database
Picsart (25.67M) is the largest AI photo editing site in the database, about 1.9 times the size of Pixelcut. PhotoRoom (24.2M) is second, about 1.8 times the size of Pixelcut. Pixlr (16.21M) is third, about 1.2 times the size of Pixelcut. Cutout.pro (8.74M) is the next AI photo editing site below Pixelcut. All four sell AI credits or subscriptions next to a free tier, so the comparison is by pattern, not by category: Pixelcut is the fourth largest AI photo editing site in the database with a full teardown.
| Site | Category | Monthly visits | Source |
|---|---|---|---|
| Picsart | AI photo editing | 25.67M | Semrush, Oct 2026 |
| PhotoRoom | AI photo editing | 24.2M | Semrush, Oct 2026 |
| Pixlr | AI photo editing | 16.21M | Semrush, Oct 2026 |
| Cutout.pro | AI photo editing | 8.74M | Semrush, Oct 2026 |
Compare Pixelcut with Picsart and PhotoRoom side by side. The full database, with working filters, is on the teardowns index.
Read the next one in your inbox
This teardown is free to read. The new ones go out by email first, one a week, with the traffic source and month attached to every number. Leave your address and the next one comes to you.