Scale Labs and CAIS' Remote Labor Index: Fable 5.2 completes 41.3% of freelance projects at professional quality, above Opus 5.5's 21.3% and GPT-6 Astra's 20.8%
First reported by The Decoder ·
AI agents can now complete professional-quality freelance work more than twice as often as previously possible.
Anthropic's Fable 5.2 model has achieved a significant milestone on the Remote Labor Index (RLI), successfully completing 41.3% of real-world freelance projects at a professional quality standard. This performance far surpasses its predecessor, Fable 5.1 (18%), and rivals like Anthropic's Opus 5.5 (21.3%) and OpenAI's GPT-6 Astra (20.8%). The RLI, developed by Scale Labs and the Center for AI Safety (CAIS), assesses AI agents' ability to complete paid freelance tasks, with Fable 5.2 demonstrating a more than 16-fold improvement since the benchmark's inception. The model shows particular strength in audio editing, architectural design, and 3D modeling, even tackling complex projects that previously stumped AI. The RLI dataset comprises 240 projects from Upwork, collectively valued at approximately $144,000, reflecting a range of professional tasks from video editing to CAD design.
The substantial leap by Fable 5.2 on the Remote Labor Index indicates a rapid advancement in AI's capability to handle complex, real-world professional tasks, not just isolated skills. This suggests a potential shift in the freelance market, where AI could increasingly automate entire projects rather than just components. The RLI's use of actual paid projects and human expert evaluation provides a more robust measure of practical AI performance, moving beyond theoretical benchmarks.
This development poses significant implications for various professional fields, from creative industries to technical design, as AI agents become more viable for outsourcing. Businesses may explore integrating these advanced AI agents for project completion, potentially impacting workflows and the demand for human freelancers in specific domains. Future developments will likely focus on closing the remaining gaps in areas like industrial design and animation, further expanding AI's potential service offerings.
AI-written summary. May contain errors.