Back to today’s briefing

Story perspectives

New AI Benchmarks Set to Revolutionize Model Evaluation

6/25/2025

26 5

Summary

Story summary

HongShan Capital Group has unveiled xbench, an innovative AI benchmark designed to assess models through real-world tasks, overcoming the shortcomings of conventional benchmarks. In tandem, LiveCodeBench Pro, crafted by Zihan Zheng and a talented team, measures AI performance against human experts. Together, these benchmarks promise to enhance AI evaluation and its adaptability in real-world scenarios.

1 of 1

Story summary

HongShan Capital Group has unveiled xbench, an innovative AI benchmark designed to assess models through real-world tasks, overcoming the shortcomings of conventional benchmarks. In tandem, LiveCodeBench Pro, crafted by Zihan Zheng and a talented team, measures AI performance against human experts. Together, these benchmarks promise to enhance AI evaluation and its adaptability in real-world scenarios.