For professional animators and VFX artists, creating non-realistic, stylized content is notoriously difficult. It often requires months of specialized, handcrafted production that can cost hundreds of thousands of dollars per minute of video. For many video artists, these costs are prohibitive, preventing many promising projects from getting off the ground.

As Mago searched for a new provider for its compute infrastructure, the team realized only Cloud GPUs from Google Cloud could reliably deliver the high-availability, enterprise-grade GPUs it needed, while being able to rapidly approve new quotas to meet demand. At the same time, Mago was drawn to the platform's cutting-edge AI offerings, with the native integration of Gemini being a decisive factor. This ability to access both powerful compute and leading AI models from one unified ecosystem was key to providing the power and intuitive creativity video artists need to achieve their best work.

“Having access to Gemini made Google Cloud a no-brainer. What other cloud providers have their own large language models?” says Mago CEO and Co-fFounder Toloza. “The fact Google Cloud is on the bleeding edge of technology means we can deliver the best services to our users.”

Mago migrated its proprietary AI architecture to Compute Engine instances running on 80GB A100 GPUs, allowing it to deliver the rich detail and frame-to-frame consistency users require for state-of-the-art video stylization.

Mago’s AI-enabled context and prompting feature uses Gemini's multimodal capabilities to analyze users’ videos and images, automatically generating accurate, detailed text descriptions of the video. These descriptions provide an optimized starting prompt for users to continue to stylize their video using Mago’s natural language interface.

“It's very difficult for users to accurately describe all the elements of a video or an image, particularly for those who don’t speak English well,” explains Alvaro Lamarche Toloza. “Having access to this prompting feature with Gemini is a lifesaver.”

This feature, combined with Gemini 2.5 Flash Image (Nano Banana) for image-to-image processing, has significantly improved the user experience, increasing the percentage of users who successfully stylize their images in 2 minutes from 71% to 93%.

As part of that image stylization process, Mago uses Gemini to translate the creator's simple requests into precise, detailed instructions, ensuring the output matches the user's aesthetic before committing the video to render. Mago has also used Gemini to develop an AI assistant to help users as they work on their projects. By analyzing the user's current work and style references, the assistant recommends optimized, ready-to-use settings to enable them to achieve their visions, reducing complexity and speeding up the production process.

Mago is now preparing to significantly expand its user base and enable many more creative projects with the launch of its open beta phase. To support this growth, Mago will use advanced cost-saving features with Google Cloud, such as dynamic scaling and spot instances. By breaking long videos into smaller pieces, Mago can utilize these cheaper resources for background jobs, reducing operating costs without compromising the quality of the final professional video.