Anthropic’s Fable 5.2 completes 41.3 per cent of real freelance projects to a standard where a paying client accepts the work. This figure comes from the Remote Labor Index, a benchmark run by Scale Labs and the Center for AI Safety.
The result represents a sharp rise from 2.5 per cent a year ago. Manus, the leading system at that time, managed only that small fraction of tasks. Fable 5.2 now handles more than 16 times as many professional jobs.
The improvement is also large compared to recent measurements. In July, Fable 5, an earlier model from the same family, led the list at roughly 16 per cent. That score was already a significant step forward, showing the best agents had quadrupled their success rate within eight months. Four months later, Fable 5.2 reached 41.3 per cent. This is roughly two and a half times the score of Fable 5 from July. Its direct predecessor, Fable 5.1, trails far behind at just under 18 per cent.
The next spots on the current leaderboard go to Anthropic’s Opus 5.5 at 21.25 per cent and OpenAI’s GPT 6 Astra at 20.83 per cent. Fable 5.2 works almost twice as successfully as these systems. Google’s strongest model in the test, Gemini Argon 4, lands in fifth place at 17.08 per cent. Older models like GPT-5.2 or Gemini 3 Pro sit below three per cent.
Strengths in audio, architecture and 3D
Scale Labs noted that Fable 5.2 makes its biggest gains in editing audio and music, designing architecture and interior spaces, and creating 3D models. The agent also handles long, complex projects that previously top-ranked models regularly failed.
Nearly 50 of Fable 5.2’s successful completions were jobs no AI agent in the test had solved before. Scale Labs says the AI can still barely handle other tasks, especially industrial product design and animations. When AI agents fail in the test, they typically deliver amateur-quality work, incomplete results, broken files, or contradictory designs.
Real jobs worth 4,000
The Remote Labor Index covers complete work assignments rather than isolated skills. The dataset includes 240 completed projects from 358 verified professionals on the Upwork platform. Together, these jobs are worth about $144,000.
A single project costs an average of $632.60 and takes a human freelancer an average of just under 29 hours. Tasks range from video editing to CAD design, game development, and data analysis. To work on these jobs, AI agents get access to a virtual workspace with common professional software like Blender, GIMP, or Audacity, along with specialised AI tools for image, audio, and video generation.
Trained human experts manually check whether an AI agent completed a job successfully. They open the submitted files in the relevant professional software and compare the AI’s output directly with the work the human freelancer delivered. CAIS’s attempts to automate this review with another AI have so far failed. Because AI reviewers can barely operate the professional software themselves, they are fooled by submissions that look polished on the surface but are technically flawed. This causes them to rate the models far too generously.
What it means
For people using AI to take on freelance work, the gap between capability and reliability has narrowed quickly. A year ago, most attempts resulted in rejected submissions. Now, nearly one in two attempts meets client standards in specific fields. This shift suggests that AI agents can increasingly handle end-to-end professional tasks in audio, 3D, and design without constant human hand-holding. However, the data also shows that success is not universal; complex industrial design and animation remain areas where current agents struggle to produce usable files.
