gpt-image is a production-oriented AI media SaaS built on top of a modern Next.js stack. While the repository started from a broader MkSaaS foundation, the current implementation is clearly focused on image generation, image editing, and video creation workflows. The application combines a marketing site, authenticated user dashboard, credit-based billing system, and provider-facing task orchestration into a single codebase that is designed for both rapid iteration and real deployment.
The product goal is straightforward: provide users with a polished web interface for generating visual content with advanced AI models, while giving the development team a maintainable platform for authentication, payments, localization, analytics, and operations. In practice, this means the project is not only a frontend experience for prompts and uploads, but also a complete SaaS system with account management, payment flows, asset handling, and background-style generation tasks.
Core Capabilities At the product level, the platform supports multiple creation modes. On the image side, the codebase currently exposes a GPT Image 2 route for text-to-image, image-to-image, and reference-guided editing. On the video side, the platform is structured to support several generation models, including GPT Image based motion workflows, Seedance, Kling, Veo, and Sora-style routes. This model abstraction makes it easier to expand the catalog without rewriting the surrounding application logic.
The billing system is credit-driven, which fits the variable cost profile of AI generation products. The repository includes pricing, premium access, checkout, customer portal, and credit management utilities, allowing the app to support both one-time purchases and subscription-oriented monetization. Stripe is used as the payment backbone, and the surrounding code indicates attention to webhook-driven account state and credit distribution.
Technical Architecture The app is built with Next.js 15 and React 19, using the App Router and locale-aware routing. Internationalization is handled through next-intl, which makes the project suitable for multilingual launches and region-specific growth. Authentication is powered by Better Auth with a Drizzle adapter, giving the codebase a typed and relatively clean account model. PostgreSQL-style persistence is managed through Drizzle ORM and migration files in src/db.
From an infrastructure perspective, the project is prepared for Cloudflare-oriented deployment through OpenNext Cloudflare and Wrangler tooling. That makes the repository notable for teams that want a server-rendered React application while still targeting an edge-friendly deployment model.
Developer Experience The repository is organized in a way that keeps product areas easy to locate. Route handlers and pages live under src/app, reusable components are grouped under src/components, business logic and provider integrations live in src/lib and src/ai, and database definitions stay in src/db. Supporting systems such as email templates, analytics providers, operational scripts, and content are also separated into dedicated folders.
For daily development, the main workflow is based on pnpm dev, pnpm build, and pnpm lint. Biome enforces formatting and consistency, while Drizzle commands handle schema generation and migrations. Overall, this codebase is best understood as a focused AI media SaaS platform: part product website, part generation console, and part monetized infrastructure layer for shipping commercial AI creative tools.
