Skip to content

QAInsights/iamspeed.dev

Repository files navigation

⚡ I am speed.

Fast.com but for LLMs.

Features

  • Live streaming output with real-time metrics
  • Time-to-first-token (TTFT) measurement
  • Throughput tracking in tokens/sec
  • AES-GCM encrypted API key storage (local only)
  • OpenAI, Anthropic, Groq, Cerebras, Fireworks AI, Mistral, OpenRouter, Google (Gemini), x.ai, z.ai, and Local (Ollama, LM Studio, etc.) provider support
  • Race Mode — run 2–3 providers in parallel on the same prompt and watch them race side-by-side on a live race track with cars, live TPS, and a Piston Cup podium. Winner is decided by TTLT (time-to-last-token) with TPS as tiebreaker; a "Fastest Start" award highlights the lowest TTFT.
  • Community Leaderboard — submit benchmark speed results (TPS, TTFT, TTLT, provider, model) to a public leaderboard to compare different configurations and APIs side-by-side. Includes provider-specific filtering and Cloudflare Turnstile bot protection.
  • Responsive, minimal UI inspired by fast.com

Quick Start

cd iamspeed.dev
npm install
npm run dev

Open http://localhost:4321, click the gear icon to configure your API key, then hit Run.

Commands

Command Description
npm run dev Start the dev server
npm run build Build for production
npm run preview Preview the production build
npm test Run unit tests (Vitest)
npm run test:e2e Run E2E tests (Playwright)

Adding a Provider

  1. Create a new adapter in src/lib/providers/ (many providers can reuse the OpenAI-compatible adapter)
  2. Register it in src/lib/providers/index.ts
  3. Add provider metadata to src/lib/config.ts

Local Models

Select Local in settings.

  • Base URL defaults to Ollama's standard: http://localhost:11434/v1
  • Common alternatives:
    • LM Studio: http://localhost:1234/v1
    • llama.cpp server: http://localhost:8080/v1
  • Click Discover models from endpoint to fetch and auto-populate available models into a select dropdown (API key is optional).
  • Click Enter model manually if you need to override the dropdown and type a custom model ID directly.
  • Your local server must allow browser CORS:
    • Ollama: OLLAMA_ORIGINS="*" ollama serve (or export the variable)
  • HTTPS/Mixed Content Note: If you access this app via HTTPS (e.g., https://iamspeed.dev), modern browsers block insecure HTTP requests to local endpoints (Mixed Content). To work around this, either run the app locally on http://localhost:4321 (recommended), or configure a secure tunnel like Cloudflare Tunnels: npx cloudflared tunnel --url http://localhost:11434 --http-host-header localhost. Note: Free ngrok tunnels are blocked because their browser warning interstitial violates browser CORS policies.
  • The app runs entirely in the browser — no proxy.

Disclaimer

"Cars", "Piston Cup", McQueen, Sally, Chick Hicks, and related character names are trademarks of Disney/Pixar. This is an unofficial fan project and is not affiliated with or endorsed by Disney.

License

MIT

About

⚡Fast.com but for LLMs

Topics

Resources

License

Stars

Watchers

Forks

Contributors