ChatGPT, Claude, Gemini and Grok are the four most widely discussed general-purpose AI assistant apps, and each provider describes its plans, models, features and data settings in its own terms, which makes a like-for-like comparison hard. This page compares the four apps as products: who makes them, where they run, which models each plan names, which features each plan lists, how usage limits work, what business versions exist and how each app handles chats for training. It is written for individuals choosing an assistant, IT and security teams deciding which app staff may use, and buyers who need the documented facts in one table. Apps are listed alphabetically by product name and order is never a ranking. Prices and plan detail are summarised here and covered in full on the plan comparison pages linked below.
Last checked: 2026-11-15.
This is an independent reference page from NHI Mgmt Group based on public provider documentation; it is not sponsored or endorsed by any provider. Apps change often, so confirm features, plans and data settings on each provider's current pages, linked in the sources list.
Key takeaways
- All four run on the web, iOS and Android. OpenAI and Anthropic list desktop apps on their plan pages, Google's plan page links a Gemini for desktop page, and xAI lists grok.com and the iOS and Android apps, with chats and subscriptions synced across them.
- Free plans name smaller models. OpenAI lists GPT-6 Luna and GPT-5 Thinking Mini on ChatGPT Free, Anthropic lists Sonnet and Haiku on Claude Free, and Google lists a Flash-Lite model and Gemini 4 Hydrogen on Gemini Free; xAI's Grok pages checked do not name the model per plan.
- Usage limits are mostly described, not counted. Anthropic and Google describe limits that refresh every 5 hours within a weekly limit, xAI describes one weekly pool for paid users since June 2026, and OpenAI describes features as limited, expanded or maximum by plan.
- Consumer chats can be used for training unless settings say otherwise. OpenAI lists opt-out on its consumer plans, Anthropic lists opt-out through its Model Improvement setting, Google trains on chats when Keep Activity is on, and xAI points users to Data Controls in settings.
- Each provider sells a business version with no training by default or stated. Anthropic lists "None by default" for Team and Enterprise, OpenAI states it does not train on business products by default, Google directs organisations to Gemini Enterprise, and xAI lists "No training" for Grok Business and Enterprise.
What are ChatGPT, Claude, Gemini and Grok?
Each is the consumer and workplace assistant app from a model provider, built on that provider's own models. The app is a separate product from the provider's developer API: plans, limits and data terms differ between the two.
- ChatGPT (OpenAI): OpenAI's assistant, sold in Free, Go, Plus and Pro consumer plans and Business, Enterprise and Edu plans for organisations. It runs on OpenAI's GPT models. See OpenAI Models.
- Claude (Anthropic): Anthropic's assistant, sold in Free, Pro, Max and Ultra consumer plans and Team and Enterprise plans. It runs on Anthropic's Claude models. See Anthropic Models.
- Gemini (Google): Google's assistant app, sold as Free plus Google AI Plus, Google AI Pro and Google AI Ultra plans, which also bundle Google storage and other Google services. It runs on Google's Gemini models. See Google Gemini Models.
- Grok (xAI): xAI's assistant, free to start, with SuperGrok, SuperGrok Heavy and X Premium paid options and Grok Business and Enterprise for organisations. It runs on xAI's Grok models. See xAI Models.
For the underlying models compared on API terms, see Frontier AI Models Compared.
The four apps at a glance
As of 2026-11-15. Sources: each provider's plan or pricing page and help documentation.
| Field | ChatGPT | Claude | Gemini | Grok |
|---|---|---|---|---|
| Provider | OpenAI | Anthropic | xAI | |
| Consumer plans named | Free, Go, Plus, Pro | Free, Pro, Max, Ultra | Free, Google AI Plus, Google AI Pro, Google AI Ultra | Free, SuperGrok, SuperGrok Heavy, X Premium |
| Organisation plans named | Business, Enterprise, Edu | Team, Enterprise | Gemini Enterprise | Grok Business, Grok Enterprise |
| Platforms stated | Web, iOS, Android, desktop app | Web, iOS, Android, desktop | Web app and mobile app; Gemini for desktop page linked | grok.com, iOS, Android |
| Models named for Free | GPT-6 Luna, GPT-5 Thinking Mini | Sonnet, Haiku | Flash-Lite; includes Gemini 4 Hydrogen | Not stated by xAI on the pages checked |
| Usage limit style | Per-feature levels (limited, expanded, maximum) | Every 5 hours and weekly | Refreshes every 5 hours until the weekly limit | One weekly pool for paid users; separate free Chat and Voice limits |
| Consumer training setting | Opt-out available | Opt-out (Model Improvement setting) | Keep Activity on: used to improve services, including training | Data Controls in settings; training terms not stated on the pages checked |
| Lowest listed paid price | Not shown on the pricing page as fetched | Pro: US$20 a month, or US$17 a month billed annually | Google AI Plus: US$4.99 a month | Not stated by xAI on the pages checked |
Which models does each app use?
As of 2026-11-15. Model names as listed on each plan page. Apps can route requests between models, so the plan page shows access, not which model answers a given message.
| Plan level | ChatGPT | Claude | Gemini | Grok |
|---|---|---|---|---|
| Free | GPT-6 Luna, GPT-5 Thinking Mini; GPT-5.6 Terra limited to Work and Codex on desktop | Sonnet, Haiku | Flash-Lite; includes Gemini 4 Hydrogen | Not stated per plan by xAI on the pages checked |
| Standard paid | Plus: GPT-6.1 Sol, GPT-6 Astra, GPT-6 Sol, GPT-6 Luna, GPT-5.6 Sol, Terra and Luna, GPT-5 Thinking Mini, legacy models | Pro: Opus, Sonnet, Haiku; Fable through usage credits | Google AI Pro: Gemini 4 Argon and Gemini 4 Hydrogen | SuperGrok: not stated per plan by xAI on the pages checked |
| Top individual | Pro: Plus models with expanded or unlimited access, plus GPT-5.6 Sol Pro | Max and Ultra: Fable, Opus, Sonnet, Haiku | Google AI Ultra: highest access to Gemini 4 Carbon, Argon and Hydrogen | SuperGrok Heavy: not stated per plan by xAI on the pages checked |
xAI's Grok overview and FAQ pages link to an announcement of grok-4.7 without stating which plans use it. Google's plan page places "Includes" model labels inconsistently between cards, so check the live page for the current mapping. For single-model facts, see the GPT-6.1 Sol, Claude Sonnet 5.5, Gemini 4 Argon and Grok 4.7 profiles.
Where can you use each app?
As of 2026-11-15.
| Surface | ChatGPT | Claude | Gemini | Grok |
|---|---|---|---|---|
| Web | Yes, all consumer plans | Yes, all plans | Yes (web app) | Yes, grok.com |
| Mobile | iOS and Android, all consumer plans | iOS and Android, all plans | Gemini mobile app; Google notes "there might be some feature differences" from web | iOS and Android; chats, settings and subscriptions sync |
| Desktop | Desktop app; ChatGPT Work limited to desktop on Free and Go | Desktop, all plans | Gemini for desktop page linked from the plan page | Not stated by xAI on the pages checked |
| Browser or office integration | Code edits on macOS listed on all consumer plans | Claude in Chrome and Microsoft 365 from Pro | Chrome auto browse from Pro; Gemini in Gmail and Vids from Plus, Docs from Pro | Connectors for email, files and calendar |
Browser and desktop agents act with the signed-in user's sessions, which raises different risks from chat. See the Computer-Use and Browser Agents guide and Personal AI Agents.
What features does each app include?
As of 2026-11-15. Plan-page wording, summarised. "From X" means the feature is listed from that plan upwards.
| Feature | ChatGPT | Claude | Gemini | Grok |
|---|---|---|---|---|
| Voice | Free included; Go and Plus expanded; Pro unlimited; voice with video from Go | Voice mode on all plans | Gemini Live on Free | Hands-free voice conversations |
| Image and video generation | Image generation limited on Free, more on Go and Plus, unlimited on Pro | Not stated on the plan page checked | Image generation and editing on Free; video generation from Plus | Image and video creation |
| Deep research | Limited on Free and Go; expanded on Plus; maximum on Pro | Research from Pro | Deep Research on Free | Not stated on the pages checked |
| Agent or task features | ChatGPT Work; scheduled tasks from Go; Dots always-on agent on Pro | Scheduled tasks from Pro; Cowork listed on Team | Tasks in Gemini and priority Agent Mode on Ultra (select countries) | Grok Build draws on the paid weekly pool |
| Coding | Codex limited on Free and Go; included on Plus; expanded on Pro | Claude Code from Pro | Agent model rate limits in Google Antigravity from Pro | Build in the shared paid allowance |
| Memory and projects | Memory limited on Free, expanded on Plus and Pro; Projects on all consumer plans | Memory on all plans; up to 5 projects on Free, unlimited from Pro | Canvas and Skills on Free | File uploads for analysis and summarisation |
| Connectors and apps | Plugins on all consumer plans; apps for deep research from Plus | Connectors on all plans | Gemini in Gmail, Docs and Vids by plan | Connectors for email, files and calendar |
Connectors give an assistant standing access to mailboxes, drives and calendars through OAuth grants. Each grant is a non-human access path that should be inventoried and reviewed; see Custom GPTs and Gems Governance and Deep Research Tools Compared for feature detail beyond this table.
How much does each app cost?
As of 2026-11-15. US list prices as shown on each provider's page; taxes and app store prices can differ. Full tier detail is on AI Chatbot Subscription Plans Compared and AI Chatbot Free Plans Compared.
| Product | Paid individual prices listed | Usage extras stated | Source |
|---|---|---|---|
| ChatGPT | Go, Plus and Pro prices not shown on the pricing page as fetched; Pro shown as "From" | Monthly API credits on Pro; free plan for verified US K-12 teachers through June 2027 | ChatGPT pricing page |
| Claude | Pro US$20 a month (US$17 billed annually), Max US$100, Ultra US$200 | Max 5x Pro usage, Ultra 10x Pro weekly usage; US$100 and US$200 a month in API credits | Claude pricing page |
| Gemini | Google AI Plus US$4.99, Google AI Pro US$19.99, Google AI Ultra from US$99.99 (US$199.99 level) | 2x, 4x, then 5x or 20x Pro limits; AI credits extend limits; storage from 400 GB to 20 TB or more | Google AI plans page |
| Grok | SuperGrok and SuperGrok Heavy prices not stated on the pages checked | Extra Usage Credits from US$5 on the web, expiring one year after purchase; Auto Top Up | Grok FAQ |
Business and enterprise versions
As of 2026-11-15. Identity and admin controls as listed on each provider's plan page.
| Product | Plans and seat rules | Identity and admin controls listed | Training statement |
|---|---|---|---|
| ChatGPT | Business from 2 users, monthly or annual; Enterprise annual through sales; Edu for universities | Not listed on the pricing page as fetched | OpenAI does not train on business products by default (data usage page) |
| Claude | Team 2 to 150 seats (Standard US$25 monthly or US$20 annual, Premium US$125 or US$100); Enterprise US$20 a seat annual plus API-rate usage, 20-seat minimum | Team: SSO, central billing; Enterprise adds SCIM, audit logs, role-based access, custom data retention, HIPAA-ready option | "None by default" for Team and Enterprise |
| Gemini | Gemini Enterprise, referenced from the Google AI plans page | Not listed on the consumer plan page checked | The plan page's data-training statements refer to Gemini Enterprise |
| Grok | Grok Business US$30 a user a month; Enterprise through sales | Business: role-based access, domain verification, custom data retention, audit and security controls; Enterprise adds custom SSO, SCIM, customer-managed encryption keys, dedicated data plane | "No training" on both; SOC 2 Type I and II listed |
SSO and SCIM decide whether leavers lose access automatically. For admin settings, see the ChatGPT Enterprise Security Guide, the Claude for Enterprise Security Guide and the Gemini Enterprise Security Guide; for Grok, xAI's business page lists the controls above.
Are my chats used for training?
As of 2026-11-15. Consumer plans only; business terms are in the previous section.
| Topic | ChatGPT | Claude | Gemini | Grok |
|---|---|---|---|---|
| Training on chats | Used for training unless you opt out ("do not train on my content" or Data Controls) | Used for training unless you opt out (Model Improvement setting, listed as "Opt-out" on the plans page); also used if the chat is flagged for safety review or you otherwise opt in | With Keep Activity on, used to develop and improve services, including training; with it off, not used for training unless you send feedback | Not stated on the pages checked; Data Controls under Profile, Settings |
| Private mode | Temporary Chat: not in history, no memory, not used for training | Incognito chats excluded from model improvement | Temporary chats not used for training; kept 72 hours | Not stated on the pages checked |
| Retention stated | Not stated on the pages checked | Feedback stored up to 5 years, de-linked from user ID | Activity auto-deleted after 18 months by default (3 or 36 months selectable); 72 hours with Keep Activity off; human-reviewed chats kept up to 3 years | Deleted chats and files leave systems within standard retention windows; deleted accounts restorable within 30 days |
| Human review | Not stated on the pages checked | Flagged conversations may be analysed for Usage Policy enforcement | A subset of chats reviewed by people, including trained service providers | Not stated on the pages checked |
Anthropic's page also states that raw connector content (such as Google Drive or MCP servers) is excluded from training unless copied into a chat. For staff guidance, see the Safe Use of AI Chatbots Guide and AI Data Retention and Deletion. Personal accounts used for work fall outside company controls; the Blocking Personal ChatGPT, Claude and Gemini Accounts at Work guide covers enforcement.
Benchmarks
Benchmarks measure models, not apps. None of the four providers publishes an app-level benchmark on the pages checked (2026-11-15), and app answers also depend on routing, system prompts, tools and plan limits. Provider-reported model scores are on the model profiles linked above, and independent leaderboards are not reproduced here. See How to Read LLM Benchmarks.
How to choose an app: factual criteria
These are criteria to test, not recommendations.
| Criterion | Facts to check |
|---|---|
| Ecosystem you already use | Gemini plans bundle Google storage and Gemini in Gmail and Docs; Claude Pro lists Microsoft 365; Grok lists email, file and calendar connectors; ChatGPT lists plugins and apps |
| Free use | Which models and features each Free plan names; Gemini lists Deep Research on Free, Claude lists Research from Pro |
| Limit behaviour | 5-hour windows (Claude, Gemini), weekly pool (Grok paid) or per-feature levels (ChatGPT) |
| Training and privacy | Which setting controls training, whether a private mode exists, and how long data is kept |
| Work use | Whether the business plan lists SSO, SCIM, audit logs and a no-training default |
| Coding and agents | Codex, Claude Code, Google Antigravity limits and Grok Build, and which plan includes each |
| Platforms | Desktop and browser integrations differ; mobile feature sets can differ from web |
An individual can trial free plans side by side. An organisation should decide on a managed business plan first, then block or allow personal accounts accordingly; see Enterprise AI Licences vs Consumer AI Tools.
Practitioner checklist
- Record which AI assistant apps staff use, on which plan, including personal accounts expensed for work.
- On any consumer plan used for work, turn off training (ChatGPT Data Controls, Claude Model Improvement, Gemini Keep Activity, Grok Data Controls) and confirm the setting after app updates.
- Prefer a business plan that lists SSO and SCIM so joiners and leavers are handled automatically, and confirm its training statement in the contract.
- Inventory connectors and plugins granted to each app and review their OAuth scopes and owners each quarter.
- Set rules for agent features (ChatGPT Work, Cowork, Tasks in Gemini, Grok Build) before enabling them, including which accounts they may act in.
- Check retention: Gemini's 18-month default auto-delete, Claude's 5-year feedback storage and Grok's 30-day account restore window may not match your policy.
- Re-check plan pages each quarter; models named per plan change often.
Standards and references
- OpenAI: ChatGPT pricing (accessed 2026-11-15)
- OpenAI: How your data is used to improve model performance (accessed 2026-11-15)
- Anthropic: Claude plans and pricing (accessed 2026-11-15)
- Anthropic Privacy Center: Is my data used for model training? (accessed 2026-11-15)
- Google: Google AI plans for Gemini (accessed 2026-11-15)
- Google: Gemini Apps Privacy Hub (accessed 2026-11-15)
- xAI: Grok overview (accessed 2026-11-15)
- xAI: Grok FAQ (accessed 2026-11-15)
- xAI: Grok for Business (accessed 2026-11-15)
Related NHI Mgmt Group resources: AI Chatbot Subscription Plans Compared, AI Chatbot Free Plans Compared, Frontier AI Models Compared, Claude Sonnet 5.5 vs GPT-6.1 Sol vs Gemini 3.8 Flash.