QuiverAI Launches Arrow 1.0, the First SVG-Native AI Model, With $8.3M From a16z

QuiverAI has opened public beta access to Arrow 1.0, the first AI model built specifically for generating structured, editable SVG graphics from text prompts and images. The launch comes alongside the announcement of an $8.3 million seed round led by Andreessen Horowitz (a16z), with participation from K Fund, JME, Mission, and notable angel investors including Webflow CEO Linda Tong, Replit CEO Amjad Masad, and Cursor's Eric Zacariason.

Why SVG-Native Matters

Most AI image generators produce pixel-based outputs — PNGs, JPEGs, or WebPs. While these look good on screen, they lack the structure needed for production design workflows. You cannot easily change colors across an entire illustration, isolate and edit individual layers, animate elements, or scale them infinitely without quality loss.

Arrow 1.0 takes a fundamentally different approach. Instead of generating pixels, it treats graphics as code, producing clean SVG markup that designers and developers can directly edit, animate, and integrate into their projects.

Key Capabilities

Text-to-SVG: Generate vector graphics from natural language prompts
Image-to-SVG: Convert raster images into editable SVG code
Natural language editing: Refine and iterate on generated SVGs with conversational commands
Clean output: Minimal control points and artifacts, well-structured layers
Complex shape support: Gradients, polygons, typography, and technical drawings

The model has already topped the SVG Arena leaderboard with a score of 1583, the first model to ever cross the 1500 threshold, outperforming general-purpose LLMs like Claude, Gemini, and GPT on vector generation tasks.

The Investment Thesis

a16z frames the investment around a critical gap in AI capabilities. As the firm's partners Yoko Li, Guido Appenzeller, and Martin Casado noted, "Code generation required models that understood syntax and semantics; design generation will require models treating visuals as structured systems."

The founding team is led by Joan Rodríguez, a researcher who previously created StarVector and pioneered reinforcement learning approaches for SVG generation (RLRF). The team combines deep research expertise with design and product engineering backgrounds.

Use Cases for Developers and Designers

Arrow 1.0 is particularly strong for logos and brand identity, icon sets and UI components, technical diagrams and floorplans, fonts and typography, and animations. For web developers, the ability to generate production-ready SVG code that integrates directly into CSS and JavaScript workflows opens new possibilities for rapid prototyping and design iteration.

Availability and Pricing

Arrow 1.0 is available now in public beta at app.quiver.ai. A free tier offers 20 SVG generations per week, with enterprise pricing available for higher volume needs. The API is documented at docs.quiver.ai for integration into existing design and development pipelines.

What's Next

QuiverAI has animation support on its roadmap, which would allow users to generate animated SVGs from prompts — a capability with significant implications for web design, motion graphics, and interactive media. As vector graphics become increasingly important in responsive web design and cross-platform applications, a dedicated AI model for the format could reshape how visual assets are created and maintained.

Source: QuiverAI