F13 emerges from stealth with $5 million raise and AI model for precise vector graphics

The company says its model can generate editable, scalable graphics for scientific diagrams, educational materials, charts and other technical visuals, with early access now open

F13 has raised $5 million to develop AI models for precise, editable vector graphics

AI startup F13 has emerged from stealth with a $5 million pre-seed round and a model designed to generate precise vector graphics, including scientific and educational diagrams where proportions, data and layout need to remain accurate.

The funding was led by Credo Ventures and Point Nine Capital, with participation from angel investors including Baobab Ventures’ Carles Reina and mainframe’s Jack Richardson.

F13 is positioning its model against general-purpose image generators, arguing that while existing systems can produce creative imagery, they are less suited to graphics that must accurately represent underlying information.

That includes charts based on specific data, scientific diagrams requiring correct proportions and visual assets that need to follow an established design system. Unlike raster images, the vector graphics generated by F13 remain scalable and editable after they are produced.

Users can provide text, images or existing graphics as inputs. F13 says the resulting visuals can then be edited and iterated rather than requiring users to regenerate an image from scratch.

Targeting technical accuracy rather than image generation alone

F13 describes its approach as “data-faithful,” with its models designed to preserve the information and specifications behind a visual.

The company says internal testing against general-purpose image and code-generation models shows that F13 outperforms models from major AI labs when generating vector graphics for clarity, style and precision. It has not provided details of the benchmark, competing models or testing methodology in the announcement.

Gregory Janik, Co-founder and CEO of F13, says the company deliberately focused on accuracy rather than building another general-purpose image model: “We didn't want to build another model that's impressively expressive and quietly wrong. We create visuals that are not only beautiful to look at, but are true to the brief, with the correct proportions, accurate data, and precise layouts.”

That distinction becomes particularly relevant for the scientific and educational diagrams F13 is targeting, where an image that looks convincing may still be unsuitable if its proportions or underlying information are incorrect.

Other initial use cases include data visualizations and charts, editorial infographics, maps and spatial graphics, and brand-consistent presentation and marketing materials.

Early access begins ahead of public launch

F13 was founded by Janik and machine learning researcher Ingimar Tomasson after the pair worked together at Knownunity, where they were integrating precise visuals into its product.

Janik has previously built a machine learning infrastructure startup and an FCA-licensed fintech, as well as working on autonomous drones at a company later acquired by Helsing. Tomasson, a mathematician by training, has published machine learning research and worked on production AI systems, including a learning assistant handling millions of daily interactions.

The wider F13 team includes machine learning researchers with previous experience at Meta and Porsche, alongside designers who have worked for companies including Versace, Gucci and Nike.

The new funding will be used to bring F13’s first model to market and expand its compute, talent and data resources. The company also plans to develop additional models for more complex design tasks.

F13 has launched to early access customers through a web-based platform and developer API. The web platform allows teams to generate, edit, manage and share graphics, while access to the models is currently available through a waitlist.

A wider public launch is planned for later in 2026.