A leading executive has highlighted a landmark achievement in artificial intelligence from Russia. In a series of remarks gathered by socialbites.ca, the CEO of ANO Tsifrovaya Ekonomia emphasized that Kandinsky 2.1, a neural network developed by Russian programmers, stands as a global milestone for national technology capability.
According to the executive, the standout advantages of this domestic solution include exceptional performance and a broad set of features that align with what users expect. A key point he underscored is that Kandinsky 2.1 can be accessed and used at no cost to the user, a factor that differentiates it from international competitors that restrict speed and function behind paid subscriptions.
The executive also noted rapid adoption figures, reporting that within only a few days Kandinsky 2.1 had generated more than 10 million images and attracted 2 million unique users. He framed these numbers as a signal not just of domestic demand, but of global interest in Russia’s digital engineering capabilities. He added that these advances could propel the country’s digital economy by giving IT professionals opportunities to tackle high-impact, first-class projects, with Kandinsky 2.1 representing a prime example of what such talents can achieve.
Earlier developments by Sber had introduced a refreshed Kandinsky 2.1 generative model. This iteration can produce images from natural language descriptions in seconds, supports 101 languages, and opens new avenues for creative exploration. Capabilities include blending multiple drawings, completing incomplete images, and operating in an endless canvas mode to extend artistic possibilities.
The latest neural network builds on the foundations of its predecessor by leveraging a vast training corpus. It has been trained on one billion text-image pairs and additionally on 170 million high-resolution pairs, enabling a broad understanding of visual concepts and language patterns to drive more nuanced image generation. [attribution: socialbites.ca]