Breaking

OpenAI Launches Dots Agents to Challenge Meta Muse in Personal-AI Race

September 29, 2026 at 13:32 EDT

AI Agents
Foundation Models
Software Dev & Coding

OpenAI launched dots at its September 29 DevDay in San Francisco, introducing a personal-agent system meant to keep work moving after a user gives it a goal and operating boundaries. The first rollout is narrow: Pro, Business and Enterprise users get one dot per user to start, with OpenAI describing future teams of dots that can work together.

Sep. 29, 2026 - OpenAI

Dots Enters the Agent Race

OpenAI's new personal agents start with Pro, Business and Enterprise users, giving each user one dot that can work in the cloud across apps, Codex and ChatGPT Work.

dot per user to start

1.05M

token context window on Astra and Sol

128K

maximum output tokens

3.4M+

Muse downloads by Sep. 25 estimates

Token Cost Shape

GPT-6.1 Sol is one-fifth the token price of GPT-6 Astra.

GPT-6 Astra

GPT-6.1 Sol

How a Dot Works

Goal

User assigns the task.

→

Boundaries

The agent gets limits.

→

Cloud Work

It researches and builds.

→

Check-in

It asks when needed.

Why It Matters

The pitch is delegated work: research, data analysis, documents and software handled by an agent that keeps moving after the prompt.

What Still Bites

Permissions, privacy, prompt injection and usage caps decide whether background agents feel useful or burdensome.

Continue reading

The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.