MT-Bench

A benchmark that evaluates advanced language models using difficult, real-world prompts and pairwise comparisons to measure instruction-following and overall response quality.

A crowdsourced benchmark where users compare responses from multiple AI models through blind pairwise voting to rank model performance.

An evaluation method that uses a large language model to assess the quality, correctness, or other attributes of AI-generated outputs.

An evaluation method that compares AI models on the same tasks, datasets, or metrics to identify differences in capabilities and performance.