LLMcompare

https://llmcompare-drab.vercel.app/

Catalog and compare frontier and open-weight LLMs on benchmarks, list prices, context windows, and speed.

Features

• Sortable model catalog with org filters and open-weight badges
• Side-by-side comparisons with shareable URLs
• Price vs performance scatter chart
• Per-model pages with specs, pricing, and benchmark scores
• Benchmark reference pages with sources
• Latest releases timeline
• Read-only REST API with OpenAPI documentation

Model data lives in src/data/ and is updated manually.

Stack

• Next.js (App Router)
• React
• TypeScript
• Tailwind CSS
• shadcn/ui
• Recharts

Local development

npm install
npm run dev

Open http://localhost:3000.

npm run build
npm start

Project layout

src/
  app/           # Pages (home, compare, models, benchmarks, releases)
  components/    # UI and feature components
  data/          # Model and benchmark datasets
  lib/           # Helpers for models, orgs, and formatting

API

The catalog is available as a read-only JSON API. Start at /api for the current API index or /api/openapi.json for the OpenAPI 3.1 document.

All versioned endpoints live under /api/v1 and return an envelope with data and meta fields. The API is public, cacheable, and CORS-enabled.

GET /api/v1/models
GET /api/v1/models/{slug}
GET /api/v1/models/{slug}/related
GET /api/v1/benchmarks
GET /api/v1/benchmarks/{id}
GET /api/v1/organizations
GET /api/v1/releases
GET /api/v1/compare?a={slug}&b={slug}
GET /api/v1/compare/{a}-vs-{b}

/models and /releases accept q, organization (or org), family, openSource, license, modality, benchmark, minScore, sort, order, page, and limit. limit is capped at 250. For example:

/api/v1/models?q=claude&openSource=false&sort=lmarena-elo&order=desc&limit=20
/api/v1/models?benchmark=swe-bench-pro&minScore=70

License

MIT