What are OpenAI dots, and what are the alternatives?

September 29, 2026 - 4 minute read

OpenAI launched dots on September 29, 2026 as ongoing agents that use connected applications and their own computers to work toward your goals. Their tasks range from research and proposals to building and testing software.

For software teams that want more control over models and infrastructure, Factory offers a model-agnostic alternative. It combines persistent development environments with configurable background tasks and planned engineering projects.

OpenAI dots and Factory cloud computers

OpenAI dots. Each dot has its own cloud computer and browser. You can inspect that computer while it works. OpenAI also supports connections you approve to other devices, including your laptop, so dots are not limited to a remote environment. Access to your own machine is a choice rather than an automatic consequence of creating a dot.

Factory. Droid Computers provide persistent environments for engineering sessions. Files, installed packages, services, and configuration remain across sessions. Factory offers managed cloud computers, which can pause when idle and resume for new work.

Factory's Bring Your Own Machine option also lets you register a cloud VM, VPS, workstation, or on-premises server. You can run delegated work in an environment your team already maintains, with its existing development setup. Your team remains responsible for maintaining and hardening that machine.

OpenAI dots and Factory background work

OpenAI dots. Dots can look for useful work between conversations through what OpenAI calls proactive research. That mode uses read-only tools over connected applications. It cannot send messages, modify application content, or control a browser or computer. Discovering a possible task and taking action on it have separate permission boundaries.

Factory. Custom Automations start sessions from explicitly configured triggers. These include schedules, selected Slack messages, GitHub events, and incoming webhooks. Webhook triggers remain in private preview. The configuration defines what the session should do, the identity it uses, and where it runs.

Use a Factory automation when you already know the event or cadence that should start engineering work. A scheduled maintenance task and a Slack-triggered bug investigation can have different instructions and approval requirements. Dots' proactive research is useful earlier in that process, when the task itself needs to be found in connected context.

Project planning and execution

OpenAI dots. The launch describes agents that can carry several projects forward while retaining context across conversations. Examples include research, proposals, launch materials, and software development. The coding example watches customer feedback, scopes improvements, builds and tests them, and returns pull requests with videos.

Factory. Missions organize bounded engineering projects around an approved plan. You work with Droid to define the features, milestones, and success criteria before execution. The orchestration layer then coordinates implementation and validation. Smaller tasks can remain ordinary Droid sessions.

For a defined software change, agree on the affected features, required checks, and review owner before starting a Mission. Dots also covers work outside engineering, such as preparing research and proposals, so the choice depends partly on which projects you want to delegate.

Models and model selection

OpenAI dots. Dots runs on GPT-6 Astra at launch. OpenAI's announcement does not describe a bring-your-own-model option.

Factory. Custom Models let the Droid CLI and desktop app use provider keys, compatible endpoints, and local models. Custom configurations are not available in the hosted web or mobile products. The supported API formats also do not guarantee that every model will work equally well with the same task and tools.

Factory Router provides automatic model selection based on quality, latency, cost, and cache state. Use Custom Models to configure a particular endpoint, or evaluate Router for managed selection. Check Router's supported configuration before choosing it for a required provider.

If your team requires a particular provider, test it with the repository's tools and validation commands in the interface you intend to use. A local configuration may need a different setup for remote work.

Pricing and usage limits

OpenAI dots. A first dot is included in the eligible Pro and Business Premium plans described at launch. Plans include an allowance for deeper work, with expanded limits during the first month. Tasks started or managed in Codex or ChatGPT Work retain their own usage limits. Enterprise access is an admin-enabled beta. Included access does not mean unlimited task execution.

Factory. Individual plans list Pro at $20 per month, Plus at $100, and Max at $200. Usage is governed by separate rolling 5-hour, 7-day, and 30-day limits. Managed Droid Computers are included from Plus. Extra Usage is prepaid credit, purchased separately from the subscription.

For organizations, Factory's Teams plan costs $60 per month plus $40 per seat, for up to ten seats with Pro limits per seat. Business and Enterprise use custom pricing and usage terms.

To budget recurring tasks, check the eligible subscription, included allowance, and charges beyond it. Track how much of the allowance your typical tasks consume on each service, including retries and review work. The monthly fee alone will not tell you the cost of that workload.

Action rules and organizational controls

OpenAI dots. Users choose connected apps and can set Custom Rules to allow actions, require approval, or block them. Activity View exposes progress, and action auto-review checks proposed actions against instructions and safety requirements. Sensitive actions such as password changes remain with the user.

Factory. Enterprise Controls define organizational boundaries for models, tools, and settings. Hard controls remain authoritative over local configuration. Session defaults serve a different purpose and can vary within the permitted boundaries.

For a software workflow, identify who can authorize a code change, who reviews it, and which tools the session may use. Keep permission to prepare a patch separate from permission to merge or deploy it, then test those boundaries with a task that should require approval.