Google DeepMind launches a multimodal knowledge agent benchmark with visual project files
AIGoogle DeepMind and Artificial Analysis launched the Multimodal Knowledge Agent (MKA) benchmark, which tests agents on 200 private visual tasks across 10 industries. Tasks use files such as schematics and scans and grade deliverables against about 70 weighted expert criteria per task.
AIWhy it matters.
The benchmark tests agents on complex visual files with expert rubrics, and its cost-per-task data shows that top scores vary widely in price.
Loading source posts…
Source details
Artificial Analysis Articles · artificialanalysis.aiPublished · added here
Original title: Multimodal Knowledge Agent (MKA) benchmark: testing knowledge work capabilities with complex visual files