OpenAI Publishes LifeVal Everyday Task Evaluation

OpenAI published LifeVal on November 2, 2026, an evaluation designed to measure how well AI models handle practical tasks from everyday life. The company says more than one billion people use ChatGPT each week and argues that a personal assistant should take actions that complete tasks, rather than only offer advice or chat responses. LifeVal tracks model performance in areas where users have sought help from AI and personal agents, and OpenAI says it has used the evaluation to track and drive improvement over the past year.

To select tasks, OpenAI drew on the American Time Use Survey and the Multinational Time Use Study, then surveyed 3,995 demographically representative respondents about where they wanted help from a personal assistant. The resulting tasks cover caregiving, health, education, household administration, personal finances, shopping and government services. Examples include finding a kindergarten after a move, coordinating flood preparation volunteers, locating a home in Luzerne County, arranging a remote game watch and planning meals for a parent with a new diabetes diagnosis.

OpenAI reports its current models scored 93.1% on the kindergarten task, 98.0% on volunteer coordination, 100% on finding a home, 96.3% on setting up a remote game watch and 90.4% on the meal and care plan. The examples describe workflows that include shortlisting schools, preparing volunteer action plans, calculating a home transfer tax, setting up a backup stream and preparing meals, reminders and questions for a care team.

OpenAI says that over the past year it trained models to become faster, more capable and more aligned in browser use, computer use and connector tasks. It says enabling these capabilities in ChatGPT expands the tasks it can help with and moves it toward an agentic personal assistant. The company says LifeVal aims to measure assistance with decisions, everyday responsibilities and getting things done, beyond academic tests of model knowledge.

OpenAI published LifeVal on November 2, 2026, an evaluation designed to measure how well AI models handle practical tasks from everyday life. The company says more than one billion people use ChatGPT each week and argues that a personal assistant should take actions that complete tasks, rather than only offer advice or chat responses. LifeVal tracks model performance in areas where users have sought help from AI and personal agents, and OpenAI says it has used the evaluation to track and drive improvement over the past year.

To select tasks, OpenAI drew on the American Time Use Survey and the Multinational Time Use Study, then surveyed 3,995 demographically representative respondents about where they wanted help from a personal assistant. The resulting tasks cover caregiving, health, education, household administration, personal finances, shopping and government services. Examples include finding a kindergarten after a move, coordinating flood preparation volunteers, locating a home in Luzerne County, arranging a remote game watch and planning meals for a parent with a new diabetes diagnosis.

OpenAI reports its current models scored 93.1% on the kindergarten task, 98.0% on volunteer coordination, 100% on finding a home, 96.3% on setting up a remote game watch and 90.4% on the meal and care plan. The examples describe workflows that include shortlisting schools, preparing volunteer action plans, calculating a home transfer tax, setting up a backup stream and preparing meals, reminders and questions for a care team.

OpenAI says that over the past year it trained models to become faster, more capable and more aligned in browser use, computer use and connector tasks. It says enabling these capabilities in ChatGPT expands the tasks it can help with and moves it toward an agentic personal assistant. The company says LifeVal aims to measure assistance with decisions, everyday responsibilities and getting things done, beyond academic tests of model knowledge.