Compare AI model responses side-by-side in real-time.

LLM Deck is a web application that lets you send the same prompt to multiple AI models simultaneously and compare their responses in real-time. Perfect for evaluating model capabilities, testing prompts, or exploring different AI behaviors.

• Multi-Model Comparison: Compare responses from OpenAI, Anthropic, and Google models side-by-side
• Real-Time Streaming: Watch responses stream in as they're generated
• Configurable Parameters: Adjust temperature and max tokens per model
• Markdown Rendering: Responses are rendered with full Markdown support
• Modern UI: Dark theme with glass morphism design

• GPT-5.2
• o3 (Reasoning)
• o4-mini (Reasoning)
• GPT-5 Mini

• Claude 4.5 Opus
• Claude 4.5 Sonnet
• Claude 4.5 Haiku

• Gemini 3 Pro
• Gemini 3 Flash
• Gemini 2.5 Pro
• Gemini 2.5 Flash

Set the following environment variables with your API keys:

You only need to set keys for the providers you want to use.

1. Open the application in your browser
2. Add or remove model columns using the "Add Model" button
3. Configure each column with your desired provider, model, temperature, and max tokens
4. Enter your prompt in the input area at the bottom
5. Press Ctrl + Enter or click "Send" to send the prompt to all models
6. Watch the responses stream in side-by-side

The application is configurable via the config object in main.ts:

const config = {
  title: "LLM Deck",
  subtitle: "Compare AI Models",
  pageTitle: "LLM Deck — Compare AI Models",
  logo: {
    imageUrl: "", // Custom logo URL
    gradient: "from-indigo-500 to-purple-600",
    shadowColor: "indigo-500/20",
  },
  theme: {
    primaryGradient: "from-indigo-500 to-purple-600",
    primaryHoverGradient: "from-indigo-400 to-purple-500",
    accentColor: "#6366f1",
  },
  inputPlaceholder: "Enter your prompt to compare responses...",
  defaultColumns: [
    { provider: "openai", model: "gpt-5.2" },
    { provider: "anthropic", model: "claude-opus-4-5-20251101" },
    { provider: "google", model: "gemini-3-pro-preview" },
  ],
  footerText: "",
};

• Framework: Hono - Lightweight web framework
• AI SDK: Vercel AI SDK - Unified interface for AI providers
• Styling: Tailwind CSS - Utility-first CSS
• Markdown: Marked - Markdown parser and compiler

MIT