The short answer
Claude is Anthropic's AI assistant for writing, long documents, research, coding and finished work such as documents, decks and spreadsheets. Since 16 September 2026, chat and Claude Cowork are one experience, so any conversation can take on a whole task. Claude Opus 5.5, released on 22 September 2026, is the main model on paid plans, while the free plan uses Claude Sonnet and Haiku. Personal plans can train on chats unless you switch off Help Improve our AI models.
| Made by | Anthropic, United States |
|---|---|
| Free plan | Yes: Sonnet and Haiku, limits every 5 hours and weekly, Docs, Slides and Design, up to 5 Projects |
| Models | Claude Opus 5.5, Sonnet 5.5 and Haiku 5.5; Claude Fable 5.1 on Max, Ultra, Team premium seats and Enterprise, or via credits on Pro |
| Best at | Careful writing and editing, long documents, handing over whole tasks, coding with Claude Code |
| Shows sources | Yes, when it searches the web or runs Research |
| Team plans | Team (2 to 150 users) and Enterprise |
| Your data | Free, Pro, Max, Ultra: used for training unless you opt out. Team and Enterprise: not used by default. |
What Claude is in November 2026
Claude is the AI assistant made by Anthropic. It started out as the chatbot people picked for careful writing, and it is now built around a different idea: you describe a finished piece of work and Claude produces it. On 16 September 2026 Anthropic began merging chat with Claude Cowork, its mode for handing over real tasks. You no longer pick a mode first. Any conversation can search the web, read your files, run code and hand back an editable document, spreadsheet or presentation.
Anthropic shipped five new models in about five weeks. Claude Fable 5.1 and Claude Mythos 5.1 arrived on 1 September, which Anthropic calls its most advanced models for coding and knowledge work. Fable 5.1 is the one offered on Claude plans. Claude Opus 5.5 followed on 22 September. Anthropic says it performs at the level of Fable 5.1 on most work while costing about 40% less to run than Opus 5. Claude Sonnet 5.5 came on 28 September as the faster, cheaper partner to Opus, and Claude Haiku 5.5 on 7 October as the small, fast model for high-volume jobs.
Which of these you can use depends on your plan. Free gets Sonnet and Haiku. Pro, Max, Ultra, Team and Enterprise get Opus, Sonnet and Haiku. Fable costs extra usage credits on Pro, is included up to half of your weekly limit on Max and Ultra, and is included on Team premium seats and Enterprise. For most people, Opus 5.5 is the right default and Fable is for the problems where you have already seen Opus struggle.
What Claude is best at, with real examples
Writing that sounds like a person
Claude's reputation rests on writing and editing. Give it a rough draft, your notes and one example of the tone you want, and ask for a version a sceptical reader would respect. It is good at keeping your meaning while cutting length, and at telling you plainly which paragraph is weak and why. Ask it to explain its edits, and you learn something each time.
Long documents and judgement across a lot of material
Contracts, research reports, policy drafts and long email threads are where Claude earns its place. Upload the document and ask focused questions: what changed between version 2 and 3, which clauses carry the most risk for us, where does the report contradict itself. Ask it to quote the exact passage behind each point so you can check it against the original.
Handing over a whole task
Since Cowork joined every conversation, the best way to use Claude is to describe the outcome rather than the steps. "What should a competitor analysis include?" gets you a list. "Research our three main competitors' pricing pages, compare them with ours in a spreadsheet and write a one-page summary of where we are cheaper" gets you two files. In Claude Desktop you can give it access to a folder so it can read, organise and create files there, and by default it asks before permanently deleting anything. Long tasks keep running in the cloud if you close your laptop, and a task started on your desktop shows up in the mobile app so you can answer its questions on the move. Scheduled tasks can run in the cloud on a timetable.
Decks, documents and mockups from any chat
Claude Docs, Claude Slides and Claude Design are artifact templates on every plan, including Free (on Enterprise they started in beta). Docs are living documents you edit together with Claude and your team, and they export to Word, PDF, Markdown or Google Docs. Slides builds a deck from your notes or reports and exports to PowerPoint or PDF, or opens in Google Slides. Design makes mockups, one-pagers, landing pages and clickable prototypes, and exports as a .zip, PDF, PowerPoint or standalone HTML. On paid plans, Claude Dashboards (beta) builds charts from connected data such as Salesforce, and every chart shows the query behind it and when it last refreshed. Team and Enterprise also get Claude Motion (beta) for animated explainers exported as MP4.
Coding
Claude Code, Anthropic's coding agent, is included from the Pro plan upwards. Anthropic reports that Opus 5.5 leads its own benchmarks for agentic coding and computer use, and its launch page cites a tester who finished a 680,000-line code migration in under a day. Those are Anthropic's own claims, so judge it on your own codebase. If you write code at work, a week on Pro is a cheap way to find out.
Where Claude falls short
- Usage limits. Every plan has them. Free resets every five hours with a weekly cap, Pro and Max have five-hour and weekly limits, and heavier models and long tasks use your allowance faster. Even Ultra, which has no five-hour limit once your account is verified, keeps a weekly limit and a separate weekly limit for Fable.
- The free plan has no Opus. Free users get Sonnet and Haiku, up to five Projects and no Claude Code, Research, Claude in Chrome or Dashboards.
- Some of the newest parts are previews. Computer use, where Claude clicks and types in apps on your screen, is in beta on Pro, Max and Ultra and only works while Claude Desktop is open. Claude Duet, the proactive agent app, is a research preview for US Pro, Max and Ultra users through a waitlist.
- It still makes mistakes. Claude is careful, but it can misread a figure or a clause. Ask for quotes and sources and check anything you will send or sign.
- It is not the everything app. If you want one assistant with built-in image creation, voice and a huge range of consumer features, see our ChatGPT vs Claude comparison and Claude vs Gemini.
How to get better results from Claude
- Ask for the finished thing. Name the output (a one-page memo, a 10-slide deck, a spreadsheet with working formulas), the material to use and how you will judge it.
- Give it your real material. Attach the report, paste the meeting notes or connect the drive where your numbers live. A deck built from your own content needs far fewer rounds of edits.
- Choose the model on purpose. Use Sonnet 5.5 or Haiku 5.5 for low-stakes jobs you will read anyway, such as summaries and rewrites, so your limit lasts. Use Opus 5.5 for judgement across lots of material. Save Fable 5.1 for the hardest problems.
- Ask for quotes and working. "Quote the clause behind each point" or "show the formula you used" makes errors easy to catch.
- Keep related work in a Project. Put the brief, style guide and reference files in one Project so every chat starts with that context. Free allows up to five Projects; Pro and above have no cap.
- Check what memory holds. Memory now works across chat and Cowork. Open Settings, then Memory, to see the topics Claude remembers, and edit or delete them. Sensitive topics stay out unless you switch on Include sensitive topics in memory.
- Turn good instructions into a skill. If you paste the same instructions every week, save them as a skill so Claude loads them when the task needs them.
Our free Claude lessons walk through these habits with short exercises, and Yuzoki kits bundle prompts by job.
Claude privacy settings and where to find them
On the personal Free, Pro, Max and Ultra plans, Anthropic can use your chats and coding sessions to train its models unless you opt out. To check, open Settings from the menu under your name, choose Privacy and look for the toggle called Help Improve our AI models. You can also go straight to claude.ai/settings/data-privacy-controls. On mobile, tap your name, then Settings and Privacy. Turning it off stops new chats and coding sessions being used for future training, but Anthropic says it does not remove data from training that has already started or from models already trained. Conversations flagged by its safety systems can still be used for trust and safety work.
Team and Enterprise plans are not used for training by default. Only Enterprise adds custom data retention, audit logs, SCIM and a Compliance API. If you work with client data, contracts or anything regulated, that difference is the main reason to use a work plan rather than a personal one.
Which Claude plan suits whom
| Plan | Pick it if | What you mainly get |
|---|---|---|
| Free | You are trying Claude or use it lightly | Sonnet and Haiku, web search, files, memory, Docs, Slides and Design, up to 5 Projects |
| Pro | You use it for real work most days | Opus 5.5, Claude Code, Research, Claude in Chrome, Microsoft 365, Dashboards (beta), unlimited Projects |
| Max | You hit Pro limits several times a week | 5x Pro usage every five hours and every week, Fable up to half the weekly limit, monthly API credits |
| Ultra | You work in long sessions every day | 10x Pro usage every week, no five-hour limit once verified, monthly API credits |
| Team | A group of 2 to 150 people shares work | Standard and Premium seats, SSO, admin controls, Cowork, Motion (beta), no training by default |
| Enterprise | Compliance and retention matter | SCIM, audit logs, custom retention, role-based access, HIPAA-ready offering |
Our honest view: Pro is the plan most professionals should start with, because it is the simplest way for one person to get Opus 5.5 and Claude Code. Measure how often you hit a limit before upgrading. If it is once a week or less, stay on Pro. Anthropic has renamed Max 20x to Ultra, with existing subscribers moved over automatically at the same price, and the 5x tier is now listed simply as Max. Ultra is sold monthly only, and the five-hour limit only disappears once Anthropic verifies your account, which it does automatically. You can check in Settings, then Usage: if there is no Current session bar, you are verified. Prices are on Anthropic's pricing page. Comparing tools? See Claude alternatives or compare tools side by side.
8 Claude prompts to copy
Replace the [placeholders]. The file-based prompts work best with the document attached rather than pasted. The weekly report prompt needs a paid plan, because scheduled tasks are not on Free, and the competitor and deck prompts go further with the higher limits of a paid plan.
Prompts to copy
Edit a draft without losing your voice
Here is my draft and one earlier piece whose tone I like. Edit the draft to be 30% shorter and clearer for [reader]. Keep my voice and every fact. Then list your five most important changes and why you made each one. Draft: [paste] Tone example: [paste]
You get: A tighter draft plus the reasoning, so you learn the pattern
Find the risk in a contract
I have attached a [supplier contract]. We are the [buyer]. List the ten clauses that carry the most risk for us, ranked. For each, quote the exact wording, explain the risk in one sentence and suggest a fairer alternative. Say if anything is unclear rather than guessing.
You get: A ranked risk list with quotes you can check
Compare two versions of a document
Attached are version [2] and version [3] of our [policy]. Make a table of every meaningful change: section, old wording, new wording and what it means in practice. Ignore formatting changes. End with the three changes our team most needs to know about.
You get: A change log you can share with the team
Hand over a competitor analysis
Research the public pricing pages of [competitor A], [competitor B] and [competitor C]. Build a spreadsheet comparing their plans with ours: [paste our plans]. Then write a one-page summary of where we are cheaper, where we are not and one pricing move worth testing. Link every source.
You get: A sourced spreadsheet and a one-page memo
Turn a report into a deck
Turn the attached [quarterly report] into a 10-slide deck for [our board]. One message per slide, a title that states the point, no more than 30 words of text per slide, and speaker notes with the detail. Put the numbers that matter most in charts.
You get: An editable deck you can export to PowerPoint
Build a weekly report that runs itself
Every Monday at 8 a.m., read the files added to [folder] in the past week and write a one-page update: what shipped, what slipped, open risks and decisions needed. Use the same headings every week and flag anything that changed since the previous report.
You get: A scheduled task that drafts your Monday update
Pressure-test a plan
Here is my plan for [project]: [paste]. Act as a sceptical [CFO]. Give me the five strongest objections, ranked by how likely they are to sink the plan. For each, suggest the evidence that would answer it. Do not soften the objections.
You get: The hard questions before someone else asks them
Explain a dataset honestly
I have attached [our customer survey export]. Tell me the three findings that matter most, with the exact numbers and the formula or filter you used for each. Then list two conclusions this data cannot support, so I do not overclaim in my presentation.
You get: Findings with the working shown, plus the limits