Mercor

OpenAI cited Mercor's APEX-Agents benchmark, which tests professional skills in law and finance, as one of the evaluations used to validate GPT-5.4's performance, and Mercor CEO Brendan Foody was quoted praising the model's professional-task capabilities.