Skip to content

Chatbot Arena (LMSYS)

Chatbot Arena is a crowdsourced open platform for evaluating LLMs through human preference.

Description

Developed by LMSYS, it uses an Elo rating system based on pairwise comparisons where humans vote for the better response from two anonymous models.

Key Metrics

  • Elo Rating: Relative skill level of the model based on thousands of matches.

Alternatives

Backlog

  • Add details on "Hard Prompts" category.