A media generation API and workflow platform for developers (Pixelflow) that can call 550+ image, video and audio models.
Segmind started as an MLOps cloud company (in 2021 its site sold "a GPU Jupyter in one minute in your browser"), and in early 2023 it switched to generative AI inference optimization, with the "voltaML authors" as its calling card. In mid-2023, as the open-source Stable Diffusion ecosystem exploded, it turned into "the fastest open-source generative model APIs", grabbing that new window. It then open-sourced distilled SDXL models (SSD-1B, Vega) on Hugging Face and gained brand exposure among developers. Its growth since has mainly come from SEO on new model names: as soon as each hot model (Seedance, MiniMax, Kling, etc.) comes out, it puts up an API page and a blog post right away, even targeting long-tail terms like "uncensored Seedance", intercepting developers and creators searching for model names and moving them onto its pay-as-you-go API and PixelFlow workflows.One-line growth story, compiled by us from public sources; speculation is marked separately below.
The main engine is SEO grabbing new model names: as soon as a hot model appears it puts up an API page and blog posts, the top search terms include elevenlabs, minimax ai and higgsfield ai free, and paid search is 0.
The lubricant is the pay-as-you-go API: after developers wire it into their own products they have to return to the console to top up and check usage, and direct visits are the largest channel.
By estimate, about half of visits are returning users going back to the console; about 7.5 in every 100 visits click through to the checkout page.
Model: Lenny's racecar growth framework. Engine = SEO grabbing new model names (as soon as each hot model appears it makes a topic page and blog post, catching developers and creators who search the model name); turbo boost = open-sourcing SSD-1B and Vega on Hugging Face early on, which gave brand exposure among developers; lubricant = the pay-as-you-go API, which brings people back to top up once it is wired into a product.
Real data: 172,900 visits, -18.27% month over month, direct as the largest channel at 42.73%, top three channel ranking (direct, organic search, referrals), organic search at 100% of search, 2,500 search terms in total, top search terms (segmind, elevenlabs, minimax ai, segmind ai, higgsfield ai free), top referrers (github.com 2.36%, thrivecart.com 0.46%, appsupertools.com 0.3%), social source YouTube, countries (US 12.3%, Spain 5.95%, India 5.28%, South Korea 3.63%, Germany 3.14%), bounce rate 39.77%, 2.53 pages per visit and an average stay of 1 minute 10 seconds come from the public Similarweb page for last month; the shares of the second and third channels are not given on the public page. The 12,894 visits (about 12,900) to the Stripe checkout page come from quietmoney checkout page data, with 0 to pay.stripe.com.
Supporting evidence: Wayback snapshots show the official site changing from an MLOps cloud and inference optimization to "the fastest open-source generative model API"; SSD-1B on Hugging Face still gets 22,000 downloads per month and has 848 likes; the blog writes "Seedance 2.0/2.5 uncensored" to grab long-tail traffic; PixelFlow got only 34 votes on Product Hunt, which shows growth does not rely on launch platforms; LinkedIn says 11-50 people.
Inference: The two channel circles are real, but how each splits is an inference: typing the URL 73,900 = returning users going back to the console 62,000 + seeing open-source models and tutorials 11,900; other channels 99,000 = searching model names 50,000 + returning users searching the brand name to log in 25,000 + traffic from GitHub, Hugging Face and YouTube 24,000. The 5% click-through after landing on a model page and 3% visit after seeing open-source models and tutorials are guesses. Semrush desktop data is only used to judge that search is the bulk of the second circle, not as a whole-site share.
| Checkout | 4–6 months ago avg | 3 months ago | 2 months ago | Last month |
|---|---|---|---|---|
| checkout.stripe.com | 0 | 0 | 0 | 12,894 |
| pay.stripe.com | 1,341 | 0 | 0 | 0 |
“Showcase 'Made with Nano Banana 2→Kling O3' labels are marketing demos, not badges on user output; no referral, share or gallery”Quote from the site
New-user source: SEO (Non-brand search). Direct 62.8%, search 35.1%, referral 2.1%; non-brand search about 21%, coming in through hot model names (elevenlabs, minimax ai, higgsfield ai free), referrals from github.com, and AI referrals from pruna.ai, Gemini and ChatGPT. Developers return to call the API and log in to the console.
Semrush source breakdown (desktop)
| Attributable new users, last month | Share |
|---|---|
| SEO | 21% |
| Other | 79% |
Direct + brand search 77%. Direct and brand search count as viral only when the site shows sharing mechanics; the largest share wins, under 5% means Other.
| Country | Share of visits |
|---|---|
| United States | 12.3% |
| Spain | 5.95% |
| India | 5.28% |
| Korea, Republic of | 3.63% |
| Germany | 3.14% |
Countries (Semrush): India 17.6%, Bangladesh 14.4%, United States 12.9%, Colombia 6.6%, France 5.8%
English
YouTube
| Source | Share |
|---|---|
| pruna.ai | 1.49 |
| Gemini | 0.78 |
| ChatGPT | 0.38 |
AI referrals (Semrush): pruna.ai 1,835 (1.49%), Gemini 960 (0.78%), ChatGPT 473 (0.38%), aragon.ai 423 (0.34%), easy-peasy.ai 332 (0.27%)
2.5K keywords · brand ~77% · organic 100% / paid 0%
| Similarweb top 5 keywords |
|---|
| segmind |
| elevenlabs |
| minimax ai |
| segmind ai |
| higgsfield ai free |
Non-brand queries: elevenlabs, minimax ai, higgsfield ai free
6 referring sites
| Site | Share of referrals |
|---|---|
| github.com | 2.36% |
| thrivecart.com | 0.46% |
| appsupertools.com | 0.3% |
| Category | Share of referrals |
|---|---|
| Programming and Developer Software | 11.72% |
| Computers Electronics and Technology - Other | 6.01% |
| Investing | 3.99% |
| Visual Arts and Design | 0.46% |
About 12,894 last month (Similarweb estimate), #263 of 606; the month before was 0. Opening checkout is not paying, so this is not revenue.
New-user source: SEO (Non-brand search). New users mostly come from need-based searches Direct 62.8%, search 35.1%, referral 2.1%; non-brand search about 21%, coming in through hot model names (elevenlabs, minimax ai, higgsfield ai free), referrals from github.com, and AI referrals from pruna.ai, Gemini and ChatGPT. Developers return to call the API and log in to the console.
Pay-as-you-go (first top-up $10); subscriptions Pro $39/month, Business $99/month, Scale $599/month; dedicated GPUs billed hourly
Dossier confidence: high. Parts 1 and 2 come from reading the site and Similarweb data; part 3 from public reporting, founder posts, Wayback snapshots, WHOIS and similar, with speculation marked.
The number: checkout visits is Similarweb's estimate of visits from the site to a checkout payment page; opening checkout is not paying, so this is not revenue. Small sites land on fixed buckets.
Data through last month. Page updated 2026-10-31.