OpenAI announced it is introducing an automatic system that will route Dots tasks to different underlying AI models, first reported by Fortune. The router is meant to reduce Dots’ computing demands, a key constraint for assistants intended to become always available, even as more powerful, constantly running agents consume more compute.
At a dinner with reporters in New York City, Adam Fry, head of ChatGPT’s consumer business, described Dots as compute-intensive. Each Dot has its own cloud computer to execute tasks using accounts and services it has been given access to.
The assistants help people read and respond to emails and find and book tickets to events. Dots are available only to people with a Pro subscription, which costs $100 per month.
OpenAI says the new task router will send different tasks to different underlying models, including GPT-6.1 Terra. Terra is smaller and uses less computing power than GPT-6 Astra, the larger model that originally powered all Dots tasks.
Musk said this week that the cost of running Grok Bot was “unsustainably high relative to revenue.” He suggested compute allocation was being throttled back, causing users to hit their limits faster.
Fry expects a breakthrough in compute efficiency to make costs fall rapidly even as model performance improves. That expected efficiency sits beside a counterpressure: more powerful, constantly running personal agents consume more compute.
OpenAI also kicked off efforts this week to rebrand data centers as compute campuses, amid a nationwide, bipartisan backlash. Community pushback could slow permits or kill facilities if benefits to local communities, including job creation and environmental protection, are not clear enough. OpenAI has promised to reduce strain on local electrical grids and not take the limited supply of power from local families.