What counts as a "Claude agent"?

Anthropic ships agent capabilities across its Claude products — for example an agentic coding tool in the Claude Code line and agent-style workflows elsewhere. The exact naming, features and packaging change, so this page keeps the comparison at category level and points to anthropic.com for the current picture.

How is a Dot different from a Claude agent?

A Dot is designed to be always-on and identity-bearing inside OpenAI's ecosystem: you name a primary Dot, it has its own cloud computer and browser, learns from feedback over time, connects to 4,000+ apps, and you reach it through ChatGPT, Slack and Teams. Anthropic's agent tooling is generally strongest where developers already work, especially in coding and API-driven workflows. Which is "better" depends on where your work lives.

Which model powers each?

Dots run on GPT-6 Astra. Anthropic's agents run on Anthropic's own Claude models. We do not restate competitor model versions here because they change; check the vendor's current documentation.

Should a developer choose Dots or Claude agents?

It depends. If you live in ChatGPT and want persistent agents that own ongoing work and reach you across channels, Dots are the natural fit. If your work is concentrated in coding and developer tooling, Anthropic's agent products are built around that context. Many teams use more than one; review consequential output either way, since agents can make mistakes.

OpenAI Dots vs Claude Agents

Comparing OpenAI Dots with Claude agents requires a little care, because they are not packaged the same way. Dots are a named, always-on product inside OpenAI's ecosystem. Anthropic's agent capabilities are spread across its Claude products and developer tooling, and their naming and scope move quickly. This page therefore compares the two at category level and links anthropic.com for the authoritative current details.

On the Dots side, everything below comes from OpenAI's launch materials. Where a fact is not published, we mark it rather than estimate.

Where each model of agency comes from

OpenAI's bet with Dots is that agents should be ambient. You start with a primary Dot and name it; it has its own cloud computer and browser, connects to 4,000+ apps, learns from your feedback over time, and works 24/7. You reach it through ChatGPT on desktop, web and mobile, and also through Slack and Teams. The agent is something you own and return to, not just a tool you invoke.

Anthropic's agent story is rooted in the developer's world. Claude Code and related capabilities are aimed at people doing technical work in code and APIs — closer to an expert collaborator in a toolchain than an always-on named teammate. That is a genuine philosophical difference, not just packaging, and it shows up in almost every comparison dimension.

Interaction and continuity

Continuity is the Dot's defining trait: OpenAI says a Dot learns from feedback over time and can carry a project forward while handling several others. Developer-oriented agent tooling tends to be session- or repository-centred, which is powerful and precise but not necessarily something that "owns" a long-running responsibility for you. If your work is a recurring process rather than a code change, that distinction is decisive. See Dots use cases for the kinds of ongoing work OpenAI describes.

What is published, and what is not

We deliberately avoid restating competitor model versions, benchmarks, prices or feature counts, because those change and are the vendor's to publish. The same discipline applies to Dots: OpenAI has not published prices for additional Dots or scaling, and the plan details you can rely on are the ones in its pricing guide. For Anthropic, treat its documentation as the source of truth.

Side by side

Dimension	OpenAI Dots	Claude agents (category)
Runs where	Its own isolated cloud computer with a browser, 24/7	Centred on Anthropic's products and developer tooling; confirm current surfaces.
Model	GPT-6 Astra	Anthropic's own Claude models; versions change — check the vendor.
Identity / credentials	Named Dot; specialist Dots have their own identity and systems access	Developer- and API-oriented access models; verify with Anthropic.
App connectivity	4,000+ apps via OpenAI's plugin ecosystem	Integrations and tool use defined by the vendor; check current docs.
Approvals / governance	Read-only research, auto-review, Custom Rules, activity view; can be paused	Vendor-specific controls; review Anthropic's published documentation.
Best fit	Ongoing, cross-channel work owned by a persistent agent	Technical workflows and coding contexts; confirm against your use case.
Primary source	OpenAI, Introducing dots	anthropic.com

The model underneath

Dots run on GPT-6 Astra, and OpenAI frames that model plus the Dot's own computer as the basis for taking action. Anthropic's agents run on Anthropic's own Claude models. For a buyer, the practical difference is less about leaderboard comparisons — which we do not repeat — and more about which ecosystem's model behaviour, tooling and data controls you already have in place.

It depends on your needs

If you want an always-on teammate that reaches you in ChatGPT and Slack and takes ownership of ongoing work, Dots are designed for that. If your priority is agentic help embedded in technical work and developer tooling, Anthropic's direction is often the closer fit. Neither is universally better; the tiebreaker is where your work already happens. And whichever you adopt, review consequential output — agents can make mistakes.

Source for Dots facts: OpenAI, Introducing dots. Independent guide; not affiliated with OpenAI or Anthropic.