Tag
Mercor hired 12 licensed CPAs to complete simplified month-end close tasks from its APEX-Accounting benchmark and found that frontier AI models now score perfectly and are ~50x cheaper than the best junior accountant, having gone from below the human average to 100% accuracy in just 18 months.
The Opus 5.5 AI model has surpassed professional human writers on a benchmark evaluating script quality, including voice, substance, pacing, and minimal AI-generated errors.
The Vals AI RSI Index indicates that the Opus 5.5 AI model has beaten the published human baseline on one of five tasks, shifting the predicted date for achieving full RSI from August 2027 to July 2027.
GPT-5.6 Sol reportedly hits the ZeroBench human baseline at pass@5 without tools, meaning at least one of five attempts succeeds on the benchmark.
This paper audits six large language models for gender stereotyping across English, Korean, Chinese, and Japanese, anchoring against human baselines. It finds that LLM stereotyping often exceeds human cross-country variation and can compound across languages, introducing a four-pattern framework to characterize such behaviors.