New AI Model Surpasses DALL-E and Midjourney on Key Benchmark — Creator Secures $30 Million in Funding

Recraft, a startup known for its mysterious image model that outperformed OpenAI’s DALL-E and Midjourney on a prominent industry benchmark last year, has successfully raised $30 million in a Series B funding round, led by Accel, which was disclosed exclusively to TechCrunch.

Other investors in this funding round include Khosla Ventures and Madrona. Headquartered in San Francisco, Recraft had previously secured $12 million in a Series A round led by Khosla in 2024. The startup reports exceeding $5 million in annual recurring revenue (ARR) and attaining 4 million users recently.

The company gained significant recognition when its model, codenamed “red_panda,” performed exceptionally well in the Artificial Analysis benchmark last year. This model, known as V3, earned its name because early users often generated adorable images of the mammal, as detailed by Recraft’s founder and CEO, Anna Veronika Dorogush.

Recraft claims to develop its models from the ground up and faces competition from image generators like Midjourney, Adobe Firefly, Stable Diffusion, and Black Forest Labs, according to Dorogush. However, Recraft’s AI is particularly skilled at generating brand-specific images, allowing companies to accurately position logos without extra editing and effortlessly create new marketing materials, such as brochures and posters, that align with established branding guidelines.

This feature addresses a prevalent gap in existing image models, as noted by Dorogush, and brings Recraft closer to design tools like Canva, which also includes an AI generator focused on branding.

Recraft is especially notable for having a sole female founder and CEO. Dorogush founded the company after years of experience in machine learning at Yandex—Russia’s equivalent of Google—as well as positions at Google and Microsoft.

Before delving into AI model development, Dorogush worked as a professional model while pursuing a math and computer science degree at one of Russia’s top universities. Although she eventually left modeling, she believes that experience taught her that simple hard work—like attending endless casting calls—was not enough.

“The most crucial lesson I learned during that time is that persistence isn’t everything,” Dorogush expressed. “Now, as we build this company, I realize that for us to succeed, we must excel in vital areas. For us, mastering model building is essential. That’s why we are fully committed to perfecting this.”