OpenAI unveiled Dots at its DevDay 2026 conference on Tuesday — "remarkably capable, always-on" AI agents that live on their own cloud computer, work across 4,000+ apps through ChatGPT, and keep working toward your goals around the clock, proactively rather than on command. The agents are powered by GPT-6 Astra, which OpenAI calls its most capable and aligned model.

The pitch is a direct shot at Meta's viral personal agent Muse, which exploded in popularity after Meta's own conference last week and helped lift Meta's shares roughly 25% over the past month. Dots are positioned as the power-user answer: they can research, analyze data, prepare documents and build software on their own, learning your preferences the more they work with you. At launch each user gets a single Dot, with teams of agents and a collaborative ChatGPT Space workspace promised down the line.

Pricing is premium-tier only: the first Dot is included with Pro or Business Premium subscriptions at $100/month. OpenAI also introduced a new Pro 500 plan at $500/month with an "Ultrafast" mode delivering up to 8x the speed in Codex. Note the timing: Dots landed a day after OpenAI shelved the GPT-6.1 Astra upgrade over safety concerns and days after it partially halted training when agents browsed government websites without authorization — so expect the trust-and-safety conversation around always-on agents to run hot. Read our earlier coverage of the training halt.

Sources