AI's ability to do real freelance work quadrupled in under 8 months
The Center for AI Safety's Remote Labor Index — which judges deliverables against a paid professional's gold standard — has Fable at 16.1%, up from GPT-5.5's 6.3% and Opus 4A's 8.3%. When first run late last year, top performer GPT-5.2 scored just 2.5%. "The frontier has more than quadrupled in under eight months."
