terminal-gpt

bringing AI Chat to Terminal

• 📦 Out of the box, multi-provider AI chat in your terminal
• 🔄 7 built-in providers: OpenAI, Anthropic, Kimi (K2.5), Together, DeepSeek, Ollama, Custom
• 🌊 Streaming response with real-time token display
• 💬 Multi-turn conversation with context memory
• ⚡️ Per-provider API Key & model configuration
• 🔩 Friendly visual interaction powered by Vue-TermUI

Supported Providers

| Provider | Default Model | Base URL | |----------|--------------|----------| | OpenAI | gpt-4o | https://api.openai.com/v1 | | Anthropic | claude-sonnet-4-20250514 | https://api.anthropic.com/v1 | | Kimi | kimi-k2.5 | https://api.moonshot.cn/v1 | | Together | Llama-3.3-70B-Instruct-Turbo | https://api.together.xyz/v1 | | DeepSeek | deepseek-chat | https://api.deepseek.com/v1 | | Ollama | llama3.2 | http://localhost:11434/v1 | | Custom | (user-defined) | (user-defined) |

All providers use the OpenAI-compatible API format. You can override Base URL and Model for any provider in Settings.

Install

pnpm i tgpt -g

Quick Start

tgpt init

Usage

Chat

Press Tab to focus input, Enter to send, ArrowDown to open Settings.

• /clear — reset conversation history
• /exit — quit tgpt
• ArrowUp / ArrowDown — browse input history

Settings

Press Tab to input, Enter to confirm, ArrowUp to save & go back to chat.

• Press P to cycle through providers
• Each provider keeps its own API Key, Base URL, and Model
• Config is saved to ~/.tgpt/config.json

Configuration

Config file location: ~/.tgpt/config.json

{
  "providerIndex": 0,          // Active provider (0-6)
  "providerKeys": {            // API keys per provider
    "OpenAI": "sk-xxx",
    "Kimi": "sk-xxx"
  },
  "providerOverrides": {       // Optional overrides
    "OpenAI": {
      "model": "gpt-4o-mini"
    }
  }
}

Tech Stack

• Vue 3 + Vue-TermUI — terminal UI framework
• Pinia — state management
• Vite — build tool
• TypeScript

License

MIT © classmatewu