OpenAI released LifeVal on November 2, 2026, a benchmark measuring whether AI models can complete everyday personal tasks end-to-end rather than provide advisory dialogue. It covers seven categories: caregiving, health, education, household administration, personal finances, shopping, and government services. Example tasks include booking a car inspection, ordering halal groceries within budget, finding childcare, apartment hunting, and accessing government forms. Each task is graded against its own checklist rather than a single correct answer. The benchmark draws on data from the American Time Use Survey, the Multinational Time Use Study, and a survey of 3,995 demographically representative respondents, plus 100 detailed task authors.
GPT-6 Astra scores 80.6% overall, with 93.1% on kindergarten search tasks and 90.4% on diabetic meal planning, up from 49.3% for GPT-5.2 in December 2025. OpenAI stated it has used LifeVal internally for approximately one year to measure and guide model improvements. OpenAI's stated rationale is that "for AGI to benefit all of humanity, it should help people with the everyday tasks that matter most to them." LifeVal is a standalone evaluation framework and does not introduce a consumer-facing product. It complements GDPVal, a separate OpenAI benchmark covering work tasks across industries.