The Claude app, explained: plans, models, Cowork and how usage limits work

Claude is no longer only a chat window. As of October 8, 2026, one Claude subscription covers chat on web, desktop and mobile; Cowork, an agent for multi-step tasks; Claude in Chrome, a browser extension that can click and type for you; and Claude Code in the terminal. All four draw on the same usage allowance. This explainer covers what each plan includes, which models you get and how the limits work. Disclosure: 9to5AI drafts stories with Claude's help (see the AI-use note), and that's another reason we stick to Anthropic's own documentation and attribute its claims.

The plans at a glance

Last verified: October 8, 2026 (US web prices; Anthropic notes mobile prices may differ)

Max is billed monthly only. Since October 7, 2026, Max and Team subscribers also get monthly API credits: $100 on Max 5x, $200 on Max 20x, and $20 or $100 per Team seat, pooled up to $500 a month. To claim them, you link a Claude Console organization. These credits work only in the API. Anthropic says they "can't be used in Claude Code or the Claude apps." Unused credits expire at the end of each billing cycle.

Which models you get

Anthropic released three models in three weeks: Claude Opus 5.5 on September 22, Sonnet 5.5 on September 28 and Haiku 5.5 on October 7, 2026. The more expensive Fable 5.1 arrived on September 1. Our guide to which Claude model to use covers the trade-offs. What matters for app users is how each plan pays for them:

• Free users get Sonnet and Haiku. The pricing page doesn't say which versions.
• Pro users can choose Opus, Sonnet or Haiku within their limits. Fable isn't included in Pro's limits: it is billed from the start through pay-as-you-go usage credits.
• Max, premium Team seats and premium seat-based Enterprise seats can spend up to 50% of their weekly limits on Fable. Fable uses up limits faster than other models. Past that cap, you either pay with usage credits or switch models.

With Opus 5.5, Anthropic said it was raising five-hour usage limits on Pro, Max, Team and seat-based Enterprise. It also added a one-time rate-limit reset that subscribers can save and use later.

Context windows

Anthropic's help center lists a 1M-token context window in chat for Fable 5.1, Opus 5.5, Opus 5, Sonnet 5.5, Sonnet 5 and Haiku 5.5 on paid plans. Older models get 500K or 200K tokens. In Cowork, Haiku 5.5 is limited to 500K. Anthropic doesn't document the Free plan's context window. If you have code execution turned on, Claude summarizes older messages as a conversation nears the limit. That lets the chat continue but uses more of your allowance. For how context differs from memory, see our explainer on context windows vs memory vs RAG.

How usage limits actually work

Anthropic doesn't publish message counts. It publishes multiples: Pro gets "at least 5x" Free's usage per session, and Max gets 5x or 20x Pro's. Two limits apply at once:

1. A five-hour session limit. It resets five hours after your session starts.
2. A weekly limit across all models, which resets at a fixed time assigned to your account. If your plan includes Fable, a separate bar tracks Fable usage.

Both appear under Settings > Usage. Usage is shared across claude.ai, Claude Desktop and Claude Code, including IDE use. A long coding session in the terminal uses the same allowance as your chats. Several factors use up limits faster: long conversations, large attachments, Research and web search, higher effort levels, artifacts and multi-step agent work. Project files are cached and count less when reused, until the cache expires after a period of inactivity.

When you hit a limit, paid plans can switch on usage credits, which are pay-as-you-go. In Claude Code, you can also continue on API credits, billed at standard API rates.

The features, one by one

Projects. Projects are workspaces with their own files and instructions. Free accounts can have up to five. Team plans add sharing. Anthropic recommends keeping project instructions short and removing unused files to save usage.

Artifacts. Since September 16, 2026, artifacts, including Anthropic's Claude Design, Slides and Docs formats, have been available on every plan, Free included. Since June 12, you can open Claude's drafts beside the chat and mark text for inline edits. Enterprise admins must switch artifacts on.

Memory. Anthropic rebuilt memory on July 10, 2026. It now stores individual, categorized entries instead of a daily summary. Since August 25, memory spans both chat and cloud Cowork. You can edit it under Topics, and a setting excludes sensitive topics unless you turn them on. Memory is available on Free.

Research. Claude's multi-step research mode is available on Pro and above. Like web search, it counts heavily against usage.

Connectors and plugins. Connectors to outside services are available on every plan, including Free. Many are built on the open Model Context Protocol (MCP). The Microsoft 365 connector gained write tools on July 7, 2026, so Claude can draft and send email, manage calendars and edit OneDrive and SharePoint files. Teams remains read-only, and setup requires admin consent. On September 25, 2026, Anthropic opened a developer portal for submitting plugins to the Claude directory.

Cowork. Cowork brings "the agentic capabilities of Claude Code to knowledge work." You describe an outcome, and Claude works through the steps to produce reports, spreadsheets and slides, including on a schedule. Cowork requires a paid plan and is available:

• in the desktop app on all paid plans;
• on web and mobile for Pro, Max and Team (Enterprise admins must enable it);
• in the Chrome side panel for Max and Team, rolling out to Pro.

Cloud sessions run "on Anthropic's servers in an isolated environment" and keep going if you close your laptop. Cloud sessions are in beta on Team and Enterprise. To work with local files, Cowork needs the desktop app open, and it can touch only folders you've connected. Since September 16, Cowork features have been available in any chat without switching modes, rolling out gradually to Pro and Max. Cowork has three approval modes:

• Manual: Cowork asks before each action.
• Auto: Claude safety-checks each action and blocks unsafe ones. Anthropic notes this mode uses more of your allowance because of the extra checks.
• Skip: no prompts and no checks. Anthropic says to use it only when you fully trust the task.

Cowork always asks before permanently deleting files.

Claude in Chrome. The extension is available on all paid plans, in Google Chrome only. Other Chromium browsers and mobile are not supported. It can read, click, type and fill forms, works across a tab group and can run scheduled browser tasks. Cowork and Claude Code can drive it too. Anthropic is blunt about the risk: the extension "is enhanced with our safety classifiers but is still risky," and "the biggest risk facing browser-using AI tools is prompt injection attacks." It says its current configuration cut attack success "to less than 0.08%" in internal testing, and that the risk "is not zero." Claude can't visit adult or known piracy sites and asks permission before opening financial sites. Anthropic recommends a separate Chrome profile with no banking, healthcare or government accounts. OpenAI's equivalent, the cloud browser in ChatGPT Work, is covered in our ChatGPT agent mode explainer.

Claude Code. Claude Code is included on Pro, Max, Team and Enterprise, and shares limits with the apps. Our Claude Code feature explainer covers what's new.

Who each plan suits

• Free: occasional writing and Q&A, with no Opus, Research or Claude Code.
• Pro: most individuals. It unlocks Opus, Research, Cowork and Claude Code. Heavy coding days will reach the shared five-hour limit.
• Max 5x / 20x: people who use Claude Code or Cowork for hours every day, or need regular Fable access without per-use charges.
• Team / Enterprise: organizations that need shared projects, admin controls and audit logs. Enterprise also bills usage at API rates.

For a head-to-head with ChatGPT on coding and writing, see ChatGPT vs Claude. For the company behind the app, read Anthropic explained.