AI outperforms accountants on speed and accuracy in bookkeeping tasks
ai models beat licensed accountants in structured bookkeeping tasks, achieving near flawless results, but still require supervision to close the books.
a study by mercor compared the performance of 12 licensed cpAs with an average of 5.5 years of experience to ai models on simplified tasks from the apex accounting benchmark. the study found that 18 months ago, the best ai models scored below the accountants' average of 37 percent, but today they solve the same tasks almost flawlessly. claude opus 5.5, fable 5.1, and gpt-6 astra achieved 61.8 percent, 61.0 percent, and 57.9 percent of grading criteria met, respectively. however, mercor notes that the study's tasks test exactly what ai does best and left out key parts of the job, such as client communication and context building.