OpenAI Releases LifeVal Benchmark for Everyday Decision Tasks
OpenAI announced LifeVal, a benchmark that measures how well AI models handle everyday tasks rather than coding or math problems.
1 source 1h ago
Reporting from 1 source: GIGAZINE.
Why it matters
OpenAI is framing everyday decision-making as a measurable AGI milestone, and LifeVal puts its own GPT line on a scored curve from 49.3% to 80.6%.
Key Facts
- OpenAI announced LifeVal, a benchmark for everyday tasks such as choosing a kindergarten after a move and organizing a volunteer group for floods.
- LifeVal tasks were selected with reference to the American Time Use Survey and the Multinational Time Use Study and adjusted using a questionnaire of 3,995 people.
- GPT-5.2 scored 49.3% on LifeVal, GPT-5.4 scored 73.0%, and GPT-6 Astra scored 80.6%.
- OpenAI said measuring how much AI helps people make decisions and meet daily responsibilities is necessary for AGI to benefit all of humanity.
- A detailed LifeVal report is available as a PDF from OpenAI.
Its tasks include choosing a kindergarten after a move, organizing a volunteer group for flood preparation, and narrowing down where to move based on conditions. OpenAI selected them with reference to the American Time Use Survey and the Multinational Time Use Study, adjusted by a questionnaire of 3,995 people. GPT-5.2 scored 49.3%, GPT-5.4 scored 73.0%, and GPT-6 Astra reached 80.6%.
The benchmark's task set was drawn from two time-use surveys and adjusted against a questionnaire of 3,995 people, so the prompts track activities people actually spend time on rather than exam-style problems. Scores climb across the GPT line: 49.3% for GPT-5.2, 73.0% for GPT-5.4, and 80.6% for GPT-6 Astra. OpenAI pairs those numbers with processing time, plotting average time per task against score to show gains that do not come at the cost of longer runs. The company ties the benchmark to its AGI work, arguing that academic evaluation alone does not capture whether a model helps people make decisions and meet daily responsibilities.
Written with AI by Yomimono from the 1 cited source below, including Japanese-language reporting where cited. See our editorial policy.
Add Yomimono as a preferred source on Google