LLM Leaderboard

LLM Leaderboard is a comparison platform that ranks AI models by capability, price, speed, and benchmark scores. It pulls together data from over 1,200 models across 23 leaderboards, covering everything from general chat performance to coding, reasoning, math, image generation, and speech-to-text. The site replaces the hunt through scattered blog posts and vendor dashboards with one searchable table of results.

The home page shows a live snapshot of top performers: GPT-6-Astra leads the overall board, followed by Claude Opus 5 and Claude Fable 5. Separate tabs break down the best models for specific tasks — coding, reasoning, math, knowledge, and open models. Pricing and runtime sections list the cheapest input costs and fastest output speeds, so you can weigh performance against budget. A benchmark library lets you drill into individual evaluations like GPQA or MTEB clustering, showing exactly how each model scored. The model directory supports filters by vendor, type, modality, and price range, and the scoring methodology page explains how the data is collected and normalized.

Developers and researchers who need to pick the right model for a project will find the most value here. Instead of reading multiple papers or clicking through API docs, you can compare Claude vs GPT vs open models side by side. The pricing data helps estimate costs before you commit to an API, and the runtime speeds give a sense of real-world latency. It is a straightforward tool for anyone who wants to make an informed choice without the noise.