Replit Launches 'Replit Design,' an AI Design Suite Powered by Claude, GPT-5, Gemini, Kimi, and GLM

Replit launched Replit Design today, a browser-based creative suite that generates apps, sites, and brand assets from natural-language prompts using multiple third-party AI models. The product is available now to all Replit users at replit.com/design and succeeds Replit's earlier Canvas feature, with existing Canvas projects migrated into the new tool.

What it does

Replit Design lets users describe a design and generate a working result without switching between separate ideation, design, and development tools. According to Replit, the product supports "the world's leading models for design," naming Claude, GPT-5, Gemini, Kimi, and GLM as available options. Replit did not specify which versions of these models are integrated, and pricing for Replit Design has not been disclosed.

The company describes a feature called "Ambient Intelligence," which surfaces suggested design variations at each step of a project that users can accept with one click. Replit claims this brings the same jump AI made in coding — from autocomplete to full app generation — to visual design work.

Built-in reference library

Replit has integrated Mobbin, a UI/UX reference library, directly into the product. Mobbin says its library is used by more than two million designers and is updated weekly; Replit states it draws from more than 600,000 real-world screens across over 1,000 apps. Notably, Replit says access to Mobbin's data inside Replit Design does not require a separate Mobbin account — a departure from Mobbin's typical subscription model.

Templates and brand systems

Rather than treating templates purely as starting points, Replit Design allows users to insert templates from a library of "hundreds" made by outside designers at any stage of a project, including mid-build. The product also supports uploadable or from-scratch "design systems" — brand rules for color, typography, and layout — that Replit says apply automatically across every screen a user generates, aiming to prevent the visual drift that occurs when moving between separate prompting, refinement, and publishing tools.

Early access and launch promotion

Replit says early-access users applied the tool to marketing assets and published apps and sites, and cites testimonials from a brand strategy agency founder and a designer identified by their social handle. These are user statements relayed by Replit and have not been independently verified.

To mark the launch, Replit is running a "Designathon" with more than $50,000 in cash and platform credits for submitted designs. The company also held a live discussion on X at 2pm PDT the day of launch.

What this means

Replit Design is a bet that the same AI-driven collapse of specialized workflows seen in coding tools can be repeated for visual design — folding ideation, template selection, reference research, and brand consistency into a single interface. The multi-model approach (Claude, GPT-5, Gemini, Kimi, GLM) suggests Replit is treating underlying model choice as a commodity layer rather than a differentiator, betting instead on workflow integration — particularly the built-in Mobbin access — as its competitive edge. The lack of published pricing and unspecified model versions make it difficult to assess costs or actual output quality; that will depend on hands-on testing rather than Replit's own launch claims.