OpenAI's LifeVal Puts AI Through Real Life, and GPT-6 Astra Nails 80.6%
OpenAI rolled out LifeVal, a benchmark that grades AI on everyday chores like childcare, budgeting and government services. GPT-6 Astra scored 80.6 percent, a big jump from the 49.3 percent GPT-5.2 managed a year ago. It signals the race is shifting from trivia to real life.
OpenAI has a new measuring stick, and it's not about trivia. The company unveiled LifeVal, a benchmark built to test how well AI handles ordinary, messy human tasks — childcare, budgeting and navigating government services among them, according to Cointelegraph.
The headline number: GPT-6 Astra scored 80.6 percent on LifeVal. A year ago, GPT-5.2 landed at 49.3 percent. That's a roughly 31-point leap in twelve months.
The benchmark's focus says plenty about where the competition is heading. Chatbots can ace bar exams and code challenges all day long, but parenting schedules, household money and bureaucratic paperwork are the tasks people actually need help with.
LifeVal effectively moves the scoreboard from lab puzzles to the kitchen table. If a model can handle a school pickup conflict and a grocery budget in one breath, that's a different pitch to buyers than raw benchmark bragging.
Expect rivals to trot out their own Astra comparisons soon enough. For now, OpenAI gets to wave the number — and the intended message is clear: the next AI race is about daily life, not demo tricks.
TopicsOpenAI · LifeVal · GPT-6 Astra · GPT-5.2 · Cointelegraph
Source: x.com
Written by Paparazzi with AI. Not financial advice.