Meta's latest multimodal model achieved top rankings across seven major AI benchmarks, including 89.5% on GPQA Diamond and 42.5% on ARC-AGI-2, marking a significant comeback in the AI race.

A MarsBit feature focuses on Chinese scholar Chen Wenhu and his team’s AI model evaluation benchmarks, including MMLU-Pro, MMMU and MMMU-Pro, explaining how their redesign of question difficulty, answer options and multimodal tasks addresses inaccuracies in traditional evaluations.