Streaming Responses
Tokens render the instant they arrive — with live Markdown and syntax-highlighted code.
v6.2.5
Works out of the box
Windows · Linux
The AI coding assistant, inside your terminal.
Fast, elegant, and built for developers. Six independent providers — start instantly on Default with no API key, or bring your own on OpenRouter, FreeModel, AeroLink, or fully local Ollama. Code Mode, Agent Mode, Assist Mode, streaming responses, and a terminal polished down to the last character.
Install Seed Code CLI with one command. The installer automatically downloads the latest stable release for your platform.
Windowsirm https://seedcode-cli.vercel.app/install.ps1 | iexcurl -fsSL https://seedcode-cli.vercel.app/install.sh | bash0+Models via OpenRouter
0+Free-tier models
0Independent providers
0Platforms, one tool
Why Seed Code
Built for the shell you live in
Streaming AI, six providers, and a terminal UI that feels first-class.
Six Independent Providers
Default with zero setup, OpenRouter, FreeModel Claude & Codex, AeroLink — or fully local and private with Ollama. Switch with /provider.
Beautiful Output
Markdown, syntax highlighting, panels, and progress states — considered down to the cursor.
Three ways to work
Code Mode, Agent Mode & /assist
Seed Code adapts to the task — plain chat, project-aware coding, or desktop control when you need it.
Code Mode
Project-aware coding in your workspace. Enable with /codemode on — persistent .seedcode project memory, an incremental project index, and permission-gated file operations, all from the terminal you already live in.
Agent & Assist Mode · /assist
The agent engine reads, edits, searches, and runs commands in your project — and Assist Mode extends it to your desktop: browser, keyboard, mouse, windows, and vision. Turn it on with /assist on; /permission controls what it may touch.
Six providers, one CLI
Start instantly. Or bring your own AI.
Providers are fully separated — each owns its API key, model, status, and history, so switching never leaks a credential. The built-in Default connection needs no API key at all.
Default
Seed Code's built-in connection and the provider a fresh install ships with. Chat immediately — no key, no signup. Configured independently from OpenRouter.
OpenRouter
The full model catalogue on your own key. Free and Pro modes — switch with /settings mode free|pro.
FreeModel Claude
Claude-family models on the FreeModel API. Live catalogue with /model auto selection.
FreeModel Codex
GPT/Codex models on the FreeModel Responses API. Same FreeModel key as Claude — configured independently.
AeroLink
Anthropic-compatible gateway on its own key. Claude-family models, fetched dynamically.
Ollama / Local
Fully local and private. No API key — uses your installed Ollama models, with a configurable host.
Seed Code ecosystem
Three ways to use Seed Code
Powerful in the terminal, light in the browser, and in your pocket on Android — pick the workflow that fits.
Wanna use the lighter version?
Seed Code Chat is the lightweight, browser-based way to experience Seed Code — free AI chat and coding with no install and no setup.
Wanna use it on the web?
A convenient web-based AI chat and coding experience directly from the browser — quick access, easy to try, no CLI setup required for the web experience.
Ecosystem
One Seed Code ecosystem. Every way you work.
Start in the terminal, continue in the browser, and take Seed Code Chat with you on Android.
Seed Code CLI
Terminal-first AI coding assistant.
- ▹ Code Mode
- ▹ /assist
- ▹ Multiple AI providers
- ▹ Streaming responses
- ▹ Ollama support
Seed Code Chat Web
Lightweight AI chat and coding in your browser. No CLI installation required.
Seed Code Chat
Your AI assistant, in your pocket. Code, learn, research, write, and chat from Android.
- ✓ AI-powered chat
- ✓ Coding assistance
- ✓ Markdown & code
- ✓ Custom providers
- ✓ Fast and lightweight
- ✓ Web + Android
Your terminal deserves better.
Install Seed Code CLI v6.2.5 and start shipping with AI — beautifully.