Meta Launches MMI Benchmark, Frontier Open Models Hit Only ~25% Cross-Modality Accuracy

Meta's Superintelligence Labs released the Modality Maturity Index (MMI), a multimodal benchmark with 893 human-written prompts spanning text, image, audio, video, and documents. Frontier open models achieved only about 25% cross-modality accuracy.

2026-11-14 ~ 2026-11-14 · 2 related posts

• Meta's MMI Benchmark: Top Open Models Get Cross-Modal Outputs Right Only ~25% of the Time — _dieuwke_ · 2026-11-14
• Meta launches Modality Maturity Index (MMI) multimodal benchmark with open leaderboard — _dieuwke_ · 2026-11-14