A fast, minimal AI chat app with multi-model support, memory, artifacts, code execution, and a credit system.

Try it at z1.chat

• Multi-model chat via OpenRouter — switch models per conversation
• Artifacts — live-rendered HTML, SVG, Mermaid diagrams, code, and documents
• Code execution via E2B sandboxed Linux environment — Python/JS with persistent state, pre-installed pandas, numpy, matplotlib, scipy, sympy, pillow; shell commands; file upload into sandbox
• Memory — automatically extracts and organizes facts from conversations (personal, preferences, projects, style, facts)
• Web search via Tavily
• File attachments — images, PDFs, documents, spreadsheets, code files
• Branching conversations — edit any message and fork from that point
• Credit system — pay-as-you-go with Alipay (ZPay), BYOK support
• Admin panel — user management, model curation, usage stats
• Auth — email/password with email verification, JWE session cookies, password reset
• i18n — English and Chinese

• Next.js 16 (App Router, Turbopack) + React 19
• PostgreSQL + Drizzle ORM
• Tailwind CSS v4 + shadcn/ui v4 (Base UI)
• Vercel AI SDK v6 + OpenRouter
• E2B for sandboxed code execution
• Tavily for web search
• jose for JWE session encryption
• nodemailer / Mailtrap for transactional email

• Node.js 20+
• PostgreSQL

npm install
cp .env.example .env.local
# fill in .env.local
npm run db:migrate
npm run dev

App runs at http://localhost:3000.