Bringing the Artificial Analysis LLM Performance Leaderboard to Hugging Face

Hugging Face Blog · 2 years ago

Artificial Analysis has launched a leaderboard on Hugging Face that evaluates over 100 serverless LLM API endpoints across three metrics: quality, speed, and price. The leaderboard tests each endpoint 8 times daily and reports performance data including throughput (tokens per second), latency (time to first token), and pricing, with results spanning six different workload configurations from ~100 to ~10k token prompts. Engineers can now compare models and providers in one place to optimize for their specific use case, whether prioritizing speed and cost over raw quality or vice versa.