Recraft, the startup behind a mysterious image model that beat OpenAI’s DALL-E and Midjourney on a respected industry benchmark last year, has raised a $30 million Series B round led by Accel, it exclusively told TechCrunch.
Other investors in the round include Khosla Ventures and Madrona. Based in San Francisco, Recraft previously raised a $12 million Series A led by Khosla in 2024. The San Francisco-based startup says it recently passed $5 million in ARR and 4 million users.
The startup caught industry attention when its model, codenamed “red_panda,” topped the Artificial Analysis benchmark last year. This was actually Recraft’s V3, which earned its name because early users kept generating images of the cute mammal, Recraft’s founder and CEO Anna Veronika Dorogush told TechCrunch.
Recraft says it builds its own models from scratch and competes with other image generators like Midjourney, Adobe Firefly, Stable Diffusion, and Black Forest Labs, Dorogush said. But Recraft’s AI particularly excels at generating images for brands. That means allowing them to place logos exactly where needed without extra editing, or easily generating new marketing materials like brochures and posters that comply with existing branding guidelines.
It’s an area where existing image models often fall short, according to Dorogush. That puts Recraft closer to competing with design tools like Canva, which also has an AI generator for branding purposes.
Recraft is also notable for having a solo female founder and CEO. Dorogush founded Recraft after years of working on machine learning at Yandex — the Russian Google competitor — along with prior stints at Google and Microsoft.
Before building AI models, Dorogush worked as a professional model while she earned a math and computer science degree at one of Russia’s top universities. She ended up leaving that line of work but says it taught her that simply working hard — like showing up to endless casting calls — wasn’t enough.
“The biggest lesson from that time was that grinding isn’t everything,” Dorogush said. “Now when building a company, I know that to succeed, we have to be excellent at what’s mission-critical. In our case, building models is very important. So we have put all the effort into excelling at this.”
