OpenAI reports higher everyday-task scores for GPT-6 Astra in new evaluation

OpenAI announced a new internal evaluation Monday that it says gives GPT-6 Astra an 80% general-performance score on everyday tasks, compared with 49.3% for GPT-5.2, CNN reported. The tests examine skills such as navigating websites, using personal context from chat history and reaching conclusions that fit a user's needs. The company surveyed around 4,000 people and asked a smaller group of about 100 users to propose tasks, supply personal context and develop grading criteria. The evaluation covers activities such as apartment hunting and caring for a loved one. Its results measure performance against those criteria; the tasks and desired outcomes can be subjective.