OpenAI Dots vs Meta Muse: How AI agents differ in features, use cases

The AI race is moving beyond chatbots, as OpenAI and Meta build agents that can perform tasks rather than simply answer questions. While OpenAI's Dots was announced during DevDay in San Francisco on September 29, 2026, Meta unveiled Muse on September 8, 2026.

Both companies are providing AI for browsers, connected tool use, and the capability to persist in achieving a user's objective. Their products target different use cases and ecosystems.

This change is also taking place at the model level. On September 29, OpenAI released GPT-6.1 Sol, a lower-cost model that performs nearly as well as Astra for coding and computer- and work-related tasks.

Meanwhile, Meta is looking to extend Muse to small-business workflows with new integrations with companies such as Shopify, QuickBooks, Canva, Slack and Stripe.

Muse and Dots move beyond simple chatbot tasks

Meta refers to Muse as a personal AI agent that can make plans, browse the web, fill out forms, send emails, and book travel. It is running on a cloud computer called Muse Secure VM, which has its own browser. Meta states that Muse will be able to continue running after the user exits the app and resume when it needs approval or changes.

OpenAI's Dots follows a similar agent-based approach. According to OpenAI's system card for GPT-6 Astra, Dots are always-on agents powered by GPT-6 Astra. Each Dot has its own cloud computer and browser, can access connected tools, and can perform repetitive tasks in services such as ChatGPT, email, SMS and Slack.

Where Meta Muse and OpenAI Dots differ

The main difference is their positioning. Originally, Muse was meant to be a personal agent, but it has since become available via the Muse app and WhatsApp. Meta states that it is meant for common users and can be utilised for individual tasks like shopping, travelling and emailing.

OpenAI's Dots are positioned around ongoing work and productivity. They can process recurring tasks and assign tasks to subagents, the company claims. That makes their published use cases more closely tied to research, software, data and workplace workflows.

Meta takes Muse into small-business work

On September 29, 2026, Meta enhanced Muse by adding Muse for Small Business. Meta says users can link apps like Asana, Box, Canva, Dropbox, Figma, QuickBooks, Notion, Shopify, Slack, Stripe, and Zoom, among other services, as well as Facebook and Instagram business accounts. Meta says users remain in control, with approval required before Muse publishes, sends or spends.

At the same time, on 28 September, Meta also announced its Enterprise Platform, which bundles its products, including the Muse agent, Meta Business Agent, Muse API, and Muse Code, for enterprises and developers.

GPT-6.1 Sol makes OpenAI's agent strategy cheaper

OpenAI's GPT-6.1 Sol is separate from Dots. According to the company's API documentation, Sol provides almost-Astra performance at a lower cost with complex coding, computers and professional use. It costs $2 per million input tokens and $10 per million output tokens, and cached input tokens cost $0.10 per million tokens.

The comparison is therefore not simply Muse versus Dots. Additionally, while OpenAI is combining its always-on Dots into a larger model-and-agent framework, Meta is constructing an agent around its consumer and business offerings. Both are working to make AI more than just a tool that follows instructions and more of a software that can plan, use tools, and carry out its intended purposes.