According to CoinMeta, OpenAI released a new evaluation called LifeVal that specifically tests whether AI can assist ordinary people with daily tasks such as finding kindergartens, grocery shopping, handling government benefits, and taking care of sick family members. To determine the test content, OpenAI referred to survey data on the daily lives and time allocations of residents in various countries and polled 3,995 people to understand the tasks they most hoped AI could help with. These tasks covered areas such as childcare, healthcare, shopping, financial management, and education. The results published by OpenAI showed that the average score for GPT-5.2 was 49.3%, while for GPT-6, it had increased to 80.6%. The official team also showcased some high-scoring cases; for example, AI received a score of 100 for helping a family find a house by selecting properties that met their budget and commuting requirements and calculating the property transaction taxes. LifeVal is still currently an internal test designed and evaluated by OpenAI itself.