Study finds Manus completed only 2.5% of freelance tasks

Technology Oct 07, 2026

A new study shows that even the most advanced AI models are still struggling with tricky freelance projects like game development, 3-D animation, and data analysis.

The top performer, an AI called Manus, managed to complete just 2.5% of tasks successfully.

Researchers identify memory and vision roadblocks

Researchers tested models like Manus, Grok 4, and Sonnet 4.5 on hands-on creative work and found their main roadblocks were poor long-term memory and limited visual skills, both pretty essential for these jobs.

Still, the researchers stated, "While absolute automation rates are low, our analysis shows that models are steadily improving and that progress on these complex tasks is measurable."

so there's hope future AIs might actually get the hang of these tasks!