Recraft V4 Pro

• 🎨 Design-focused image model emphasizing typography and brand layouts
• 🆕 V4 rebuilt around "design taste" composition and material realism
• 🔧 Generates editable vector output directly from prompts
• 📝 Integrated text rendering where typography interacts with the scene
• 🏢 Built for brand, marketing, and print-ready design work
• 📏 Pro tier targets high-resolution, high-fidelity output
• 🖼️ Available within Recraft Studio alongside other models
• 💬 Released February 2026 by London-based Recraft

Recraft

Recraft is a London-based startup specializing in generative AI for visual content creation. The company develops text-to-image models and offers Recraft Studio, a web-based workspace for creating and editing images, vectors, and mockups. Recraft has built a…

Recraft V4 Pro is the high-resolution tier of London-based Recraft's V4 image family, released in February 2026 and tuned for professional design and branding work. It shares the same engine as the standard Recraft V4 — including its "design taste," prompt accuracy, and creative capabilities — but targets higher resolution for print-ready assets, large-format work, and detailed compositions. Recraft describes Pro delivering finer detail, cleaner geometry, and improved realism for high-fidelity applications.

The V4 generation is positioned by Recraft as a rebuild focused on what it calls "design taste": deliberate decisions about composition, lighting, color relationships, and material realism. V4 emphasizes integrated text rendering, where typography interacts with the scene rather than sitting on top as a decorative layer.

A defining feature across the family is native vector generation: Recraft produces editable vector files with structured layers and clean paths directly from a prompt, without a separate tracing or conversion step.

V4 Pro supports multiple aspect ratios and longer prompts, making it suited to posters, packaging, logos, icons, and design-system components. It is available within Recraft Studio alongside Recraft's other text-to-image and editing models.