The short answer
Grok vs Claude: pick Anthropic's Claude for writing, editing, long documents, coding and finished files such as decks and documents, with Claude Opus 5.5 on Pro since 22 September 2026 and Sonnet 5.5 on the free plan. Pick Grok, from SpaceX Super Intelligence (formerly xAI), for live search of X posts, image and video generation, or The X Subscription bundle with Cursor Pro. For most work, Claude is the better choice.
| Grok | Claude | |
|---|---|---|
| Made by | SpaceX Super Intelligence LLC (formerly xAI), United States | Anthropic, United States |
| Free plan | Daily access with limited chat and agent requests, Cursor starter allowance | Sonnet 5.5 and Haiku 5.5, web search, files, Docs, Slides and Design, 5-hour and weekly limits |
| Top model | Grok 4.7 (consumer model per tier not named) | Claude Opus 5.5 on Pro and above; Fable 5.1 via credits on Pro |
| Live X search | Yes, with X Search on | No; web search on every plan |
| Finished files | Writing canvas; no deck or document export listed | Docs, Slides and Design on every plan; Dashboards from Pro |
| Coding | Cursor in the bundle, Grok Build | Claude Code from Pro |
| Images and video | Grok Imagine; 1080p 15-second videos on Super | Not an image or video generator; Motion (beta) on Team and Enterprise |
| Personal data | Two switches: grok.com or app, and X | One switch: Help Improve our AI models |
| Team plans | Teams and Enterprise with SSO and SOC 2 | Team (2 to 150 users) and Enterprise |
| Our pick for | Social, comms, media and Cursor users | Writers, analysts, managers, developers |
Grok vs Claude in November 2026: two very different assistants
Grok and Claude are rarely bought for the same reason, which makes the choice easier than it looks.
Grok is made by SpaceX Super Intelligence LLC (SpaceXSI), the company formerly called xAI, which SpaceX acquired in February 2026. Its flagship model is Grok 4.7, released on 21 September 2026. Since 12 October Grok is sold as part of The X Subscription, which bundles Grok chat, voice, image and video generation, Grok Bot, Cursor and X Premium in Free, Plus, Premium, Super and Ultra tiers. SpaceXSI says Premium and higher get the "latest Grok models" but does not name the model on Free or Plus.
Claude is made by Anthropic. Its current models are Claude Opus 5.5 (released 22 September 2026, on Pro and above), Claude Sonnet 5.5 (28 September) and Claude Haiku 5.5 (7 October), with Sonnet and Haiku on the free plan. Claude Fable 5.1, released on 1 September, is available through usage credits on Pro and within limits on Max and Ultra. Since September, chat and Claude Cowork have been merging into one experience, rolling out on Pro, Max and Ultra, so a conversation can take on a whole task and hand back editable files.
In short: Grok is built around live information from X, media generation and a bundle with Cursor. Claude is built around careful writing, long documents and finished work. Anthropic names the model on every plan, and SpaceXSI does not.
Writing and editing: Claude wins
Claude is the stronger writer and editor. It is our pick for careful drafting, tone and structural editing, and it handles long material well, such as a 40-page report you need to tighten or a contract you need summarised clause by clause. Because every chat can now produce files, you can ask for the edited version as a Word-ready document through Claude Docs, which exports to Word, PDF, Markdown and Google Docs on every plan including Free.
Grok offers a canvas for writing, memory, custom instructions and more than 30 languages, which is enough for posts and quick drafts. For anything where the wording matters, Claude wins.
Live news and what people are saying on X: Grok wins
With X Search on, Grok searches public X posts as well as the web, so it can report what people are saying about a launch, a policy change or a breaking story as it happens. No other major assistant can do that with X. Grok has real-time knowledge only when its search tools are on, so switch them on for anything current.
Claude includes web search on every plan, including Free, and Research for deeper reports from Pro upwards. That is good for checking facts, but it cannot read the live conversation on X the way Grok can. Remember that answers built on X posts are only as reliable as those posts: ask Grok to link every post it uses.
Long documents, research and analysis: Claude wins
For judgement across a lot of material, Claude is the better tool. Upload several PDFs, a spreadsheet and meeting notes, and ask what they agree on, where they conflict and what is missing. Claude can run code to analyse files on every plan, Research on Pro and above reads many sources for a longer report, and Claude Dashboards (beta, Pro and above) builds charts from connected data such as Salesforce, showing each chart's query and last refresh.
Grok analyses files and PDFs, and its multi-agent mode, tied to SuperGrok, sends several agents at a hard question in parallel and merges the result into one cited answer. That is useful for a single hard question. For working through your own documents, Claude wins. Our best AI for research guide covers the specialists.
Finished files: decks, documents and dashboards: Claude wins
Since 16 September 2026, Claude Design, Claude Slides and Claude Docs are on all plans including Free. Slides export to PowerPoint, PDF and Google Slides, and Design exports to PDF, PowerPoint, HTML and zip. On Pro, Max and Ultra, Cowork tasks can keep running in the cloud with your computer off, and Claude in Chrome and a Microsoft 365 connection are included from Pro.
SpaceXSI's Grok pages list a writing canvas and file analysis, but no deck, document or dashboard export like Claude's. If the output of your work is a file someone else opens, Claude wins clearly.
Coding: Claude wins for most people, Grok on bundle value
Claude Code, Anthropic's coding agent, is included from Pro upwards. Anthropic describes Fable 5.1 as one of its most advanced models for coding and knowledge work, and Max, Ultra and Team include monthly API credits for building on the Claude Platform.
Grok's case is the bundle. SpaceXAI, as the company was then called, acquired Cursor in August 2026, and Grok Premium includes Cursor Pro, while even Grok Free has a starter allowance for Cursor's coding agents. Grok 4.7 is available in Cursor and in Grok Build. Third-party models in Cursor are billed separately at API list prices. For quality and a single subscription, Claude wins. If you want Cursor and X Premium anyway, Grok Premium is the cheaper route to all three.
Images and video: Grok wins
Grok generates images and video with its Grok Imagine models, and the Super tier lists 1080p videos up to 15 seconds. Claude is not sold as an image or video generator: its creative tools are Claude Design for mockups and layouts, Slides, Docs, and Claude Motion (beta on Team and Enterprise) for animated explainers exported as MP4. For pictures and clips, Grok wins.
Privacy and your data: Claude wins
Both use personal chats for training by default, so look at how much you must change. On Claude's personal plans, go to Settings, then Privacy, and switch off Help Improve our AI models. Turning it off stops future use of new chats and coding sessions, not training already done. That is one switch, and Claude is not connected to a social network.
Grok needs two. On grok.com it is Settings, then Data, then Improve the Model, and in the apps Settings, then Data Controls, then Improve the model. On X, public posts and Grok interactions are shared for training by default until you go to Privacy and safety, then Data sharing and personalization, then Grok & Third-party Collaborators. Grok's Private Chat, behind the ghost icon, is not included in model training and is deleted within 30 days.
On company plans, Claude Team and Enterprise are not used for training by default, and Enterprise adds custom data retention and audit logs. SpaceXSI says business and enterprise content is not used to improve models, and Grok Teams adds SSO, team-wide privacy settings, custom chat retention and SOC 2. Both are workable for business; Claude's single personal switch wins it.
Price and value: Claude for work, Grok for the X bundle
Free: Claude, for most work. Its free plan includes Sonnet 5.5 and Haiku 5.5, web search, file creation, code execution, memory, connectors, Docs, Slides and Design, with limits that reset every five hours and a weekly cap. Grok Free gives daily access with limited chat and agent requests plus a Cursor starter allowance.
Paid: Claude Pro is the better single AI subscription for professional work, because it adds Opus 5.5, Claude Code, Research and Claude in Chrome. Grok Premium is better value only if you also want X Premium+ and Cursor Pro. Every Claude plan has usage limits, and even Ultra keeps a weekly limit, so heavy users should check them. See Anthropic's pricing page and SpaceXSI's pricing page.
Same task, both tools: a customer-feedback briefing
Here is how each tool approaches one realistic job, based on what Anthropic and SpaceXSI document. It describes the products, not test results. The task: your product team wants a briefing on what customers think of your latest release, using 30 support tickets you exported and the public reaction online.
How Claude handles it
You upload the ticket export and ask Claude to group the tickets into themes, count them, quote one example per theme and flag anything that suggests a bug. It runs code on the file to get the counts right, uses web search for public coverage, and hands the briefing back as a Claude Docs document or a short Slides deck you can export to PowerPoint. The weak point is X: it cannot read the live conversation there.
How Grok handles it
You ask Grok, with X Search on, what people are posting about the release, grouped into praise, complaints and feature requests with links. You can upload the ticket export too and ask for themes. The weak points are the file work and the output: SpaceXSI lists no deck or document export, so you copy the result into your own tools.
Verdict for this task: Claude, because the tickets are the stronger evidence and the briefing is a file. Add Grok only if the public reaction on X is a large part of the story.
Who should pick Grok and who should pick Claude
- Pick Claude if your work is writing, editing, analysing documents or producing decks, documents and dashboards. It is the better default for analysts, consultants, lawyers, marketers and managers.
- Pick Claude Pro if you code: Claude Code and Opus 5.5 come with it.
- Pick Grok if your job depends on live conversation on X: social media, communications, journalism and community roles.
- Pick Grok Premium if you want X Premium+ and Cursor Pro anyway, or you make a lot of images and short videos.
- Not for you: skip Grok for confidential documents on a personal account, and skip Claude if image or video generation is the main job.
See also ChatGPT vs Grok, ChatGPT vs Claude and Claude alternatives.
Prompts to copy
Tickets to themes
Here is an export of [number] support tickets about [release]: [attach file]. Group them into at most six themes, count the tickets in each, quote one short example per theme and flag any ticket that looks like a bug. Use code to count, and show the counts as a table.
You get: Accurate, counted themes with evidence and bug flags
Reaction on X
Search X for posts about [product or release] since [date]. Group what people say into praise, complaints and feature requests, link two posts for each group, and tell me which accounts look like real customers and which do not.
You get: A linked snapshot of the public reaction with a reliability check
Edit for a busy reader
Edit [paste document] for [audience] who will read it on a phone. Cut it by a third, put the decision or ask in the first two sentences, keep every number, and list the changes you made at the end so I can check them.
You get: A tighter version plus a change list you can review