DeepMind's MKA benchmark exposes big gaps among top models
Artificial Analysis launched Google DeepMind's Multimodal Knowledge Agent benchmark; top seven models cluster at 95-97% on baseline Regression criteria but spread widely from 28% to 57% on harder Climb items involving blueprints, schematics, and clinical documents.
2026-11-11 ~ 2026-11-11 · 4 related posts
- MKA deep dive: top models cluster at 95-97% on basics but split 28-57% on hard tasks — ArtificialAnlys · 2026-11-11
- DeepMind's MKA benchmark splits frontier agents: 95-97% on basics, 28-57% on hard tasks — ArtificialAnlys · 2026-11-11
- MKA rubric design: top seven models cluster at 95–97% on Regression but span 28–57% on Climb — ArtificialAnlys · 2026-11-11
1 near-duplicate retellings: ArtificialAnlys