Product review
Grok review: xAI's AI assistant, SuperGrok plans and API
Grok is xAI's AI assistant for web, iOS and Android, with live web and X search, image and video generation, voice and a paid SuperGrok tier. Consumer prices are unconfirmed; API prices are published. We have not tested it yet.
The short answer
Grok is xAI's AI assistant on web, iOS and Android, with chat, live web and X search, image and video generation, voice and file analysis. We could not read xAI's consumer price pages, so SuperGrok prices are unconfirmed. The API lists per-token prices. We have not tested it. Features are vendor-reported.
What is our quick verdict on Grok?
A broad general assistant with live search, image and video generation, voice and a multi-agent mode, plus a developer API with published per-token prices and a stated SOC 2 Type 2 position. The consumer plan prices and consumer data-use terms are what we could not confirm from xAI itself. Editorial interpretation
Intended for: Individuals who want one assistant for chat, search and media generation, and developers who want an API with long context windows and published prices. Editorial interpretation
Look elsewhere if: you need confirmed consumer data-use terms before pasting in work material, want a desktop-first writing assistant (see Claude), or want search answers built around sources (see Perplexity). Editorial interpretation
Grok at a glance
| Item | Detail |
|---|---|
| Editor's rating | 7.5/10 |
| Strongest area | Workflow fit, 7/10 |
| Weakest area | Support, 4/10 |
| Best for | Individuals who want one assistant for chat, search and media generation, and developers who want an API with long context windows and published prices. Editorial interpretation*. |
| Look elsewhere if | you need confirmed consumer data-use terms before pasting in work material, want a desktop-first writing assistant (see Claude), or want search answers built around sources (see Perplexity). Editorial interpretation*. |
| Pricing | See the pricing section |
| Last updated | 2026-10-27 |
What is the evaluation status of Grok?
This review is based on xAI's developer documentation and marketing copy supplied to us. We have not run hands-on tests yet. A tested score will be added only after the planned tests are run.
| Item | Detail |
|---|---|
| Pricing | Partial: API prices read from xAI's docs (checked 2026-10-27); consumer plan prices unverified, third-party only |
| Security | Unverified: vendor statements only; see Security evidence |
| Hands-on tests | None run |
The six areas we plan to test. The editor's rating below is separate from these planned tests.
Capability and output quality: editor's rating 6.5/10, no hands-on test yet
- Known so far: xAI's docs list text generation, function calling, voice (text to speech, speech to text, real-time), image and video generation and a coding model 2. Context windows of 256k to 1M tokens are listed for the API models 1. We found no independent benchmark we could cite; unverified.
- Planned test: Run five fixed tasks (a summary, a draft, a rewrite, a data question and a cited research question) and score accuracy, editing effort and citation correctness.
Reliability and consistency: editor's rating 4.5/10, no hands-on test yet
- Known so far: API requests are rate limited per model by spend tier, and exceeding a limit returns HTTP 429 4. No uptime figure, SLA or consumer usage-limit table was found; unverified.
- Planned test: Send the same prompts three times and compare answers; log refusals, errors and limit messages.
Workflow fit and integrations: editor's rating 7/10, no hands-on test yet
- Known so far: The vendor copy lists web, iOS and Android apps, memory across chats, Canvas for writing, custom instructions and shareable conversations 5. The API adds a Batch API, structured outputs and web search 2. Vendor-reported.
- Planned test: Use Canvas, memory and a shared conversation in a real writing task, and call the API once with web search.
Pricing and value: editor's rating 5.5/10, no hands-on test yet
- Known so far: API prices are published per model 1. Consumer tiers (free, SuperGrok and above) were reported by third-party sites only 8; xAI's own plan page returned no prices to us 5.
- Planned test: Read the live plan page in a browser, then compute cost per task on the API and on a paid plan.
Privacy and security evidence: editor's rating 4.5/10, no hands-on test yet
- Known so far: The API security FAQ states SOC 2 Type 2, no training on API data without permission and 30-day default retention 3. Consumer data-use terms were not read. See Security evidence.
- Planned test: Read the consumer privacy policy and terms in a browser, check the training setting and delete a chat, and request the SOC 2 report.
Support and documentation: editor's rating 4/10, no hands-on test yet
- Known so far: A public developer documentation site with an API reference exists 2. Enterprise capacity is routed to a sales address 4. No consumer support channel, response times or SLA were found; unverified.
- Planned test: Send one support question and record the reply time.
What is Grok?
Grok is an AI assistant from xAI. It is used in a chat interface on the web and in mobile apps, and through an API for developers 25. A separate Grok experience is built into X; its terms are covered by X's own help pages 6.
What do independent users report about Grok?
We found no independent test or user study we could cite and verify today. Third-party pricing sites describe the plan tiers 8; they are not xAI and were not confirmed.
What does xAI say Grok is?
The marketing copy supplied to us describes Grok as an assistant with chat, a multi-agent mode that runs parallel agents and returns a merged, cited answer with auditable reasoning, search of the live web and X with citations, and Imagine for text-to-image and text-to-video. Vendor claim
Where does Grok work?
Web, iOS and Android, per the vendor copy 5. Developers use the API at api.x.ai, with a US regional endpoint at us.api.x.ai for US-only handling 3. Vendor claim
How is Grok deployed?
Cloud service only; we found no self-hosted or on-device option for the assistant or the API. Enterprise accounts can turn on Zero Data Retention for the API, which disables features that need stored data, such as the stateful Responses API, Files and Collections and the Batch API 3.
What limits and constraints does Grok state?
API rate limits depend on cumulative spend, with tiers starting at $50 (Tier 1), $250, $1,000 and $5,000, and HTTP 429 when exceeded 4. The vendor copy says Imagine makes images up to 2K and videos up to 15 seconds at 720p, and that SuperGrok has higher limits. No consumer message limits were published on pages we could read; unverified.
What do we know about Grok data, privacy and security?
API: xAI's security FAQ says it is SOC 2 Type 2 compliant and points customers with a signed NDA to its Trust Center, never trains on API inputs or outputs without explicit permission, and keeps requests and responses for 30 days for abuse auditing before deleting them 3. It offers an enterprise Zero Data Retention option and says it does not recommend it for most customers 3. The default endpoint does not guarantee a processing region 3. The page does not claim HIPAA compliance and points to a BAA questionnaire 3. It makes no GDPR or ISO statement 3.
Consumer app and X: We could not read xAI's privacy policy or terms (both returned 403). X's help page, read only through a search summary, says users can control whether their public data and their Grok interactions are used to train xAI models, and third-party guides describe an opt-out under Data Controls that applies going forward 6. Because Grok is tied to X, public posts are a separate data stream with their own terms. Treat chat content as potentially used for training and review until you read the policy yourself. Status: unverified. See Security evidence.
What are the key features of Grok?
All items in this list are vendor-reported from xAI marketing copy supplied to us 5, except where a docs page is cited.
- Chat, plus a multi-agent mode with parallel agents and a merged, cited answer 5
- Search of the live web and X with citations; the API's Web Search tool is documented 57
- Imagine: text-to-image and text-to-video, up to 2K images and 15-second videos at 720p, edited with follow-up prompts 5
- Low-latency voice conversations; the API lists text to speech, speech to text and real-time voice 25
- File and PDF analysis and vision 5
- 30+ languages, memory across chats, Canvas for writing, custom instructions and shareable conversations 5
How much does Grok cost?
Consumer plans: we could not read xAI's price page on 2026-10-27, so these are not confirmed. The vendor copy says a SuperGrok tier adds higher limits, priority access and multi-agent mode 5. Third-party sites report the following, which you should confirm at grok.com before buying 8:
| Plan | Reported monthly price | Status |
|---|---|---|
| Free | $0, limited usage | Unverified |
| SuperGrok Lite | $10 | Unverified |
| SuperGrok | $30 | Unverified |
| SuperGrok Heavy | $300 | Unverified |
API prices, checked 2026-10-27 on xAI's docs, per million tokens in USD for prompts under 200,000 tokens 1:
| Model | Context | Input | Output |
|---|---|---|---|
| grok-4.7 | 500k | $2.00 | $6.00 |
| grok-4.3 | 1M | $1.25 | $2.50 |
| grok-4.20 multi-agent | 1M | $1.25 | $2.50 |
| grok-build-0.1 | 256k | $1.00 | $2.00 |
When a prompt reaches 200,000 tokens, the higher rate applies to all tokens in that request, for example $4.00 input and $12.00 output for grok-4.7 1. Cached input is cheaper. Web search tool pricing was not on the page we read 7.
What limits should you know about?
- Consumer plan prices and limits could not be confirmed from xAI.
- Consumer training and retention terms were not read; the API terms differ from the consumer terms.
- The default API endpoint does not guarantee a processing region 3.
- Search and media claims are vendor-reported and untested.
- Security claims unverified.
How do you get started with Grok?
Open grok.com or the iOS or Android app and sign in; the page shows Sign in and Sign up options 5. Before using work material, open Data Controls and read the privacy policy. Developers create an API key in the xAI Console and start with the models page 12.
How will we test Grok?
We will run the same five real tasks we use for every productivity tool, with the same prompts in Grok, Claude and Perplexity, and record accuracy, editing effort, citation correctness, limit messages and time. Measured results will be kept separate from editorial judgment.
What are the alternatives to Grok?
Editorial and unranked. Claude is a general assistant on web, desktop and mobile with published consumer training controls. Perplexity is built around cited search answers. See Productivity for the other tools in the category. Editorial interpretation
What is the final verdict on Grok?
Grok has a wide feature list and a clearly documented, priced API. Consumer prices, consumer data use and all quality claims remain unconfirmed. Not tested yet. Editorial interpretation
What is the public record for Grok?
| Date | Event | Source |
|---|---|---|
| 2026-10-27 | API models, prices, rate limits and security FAQ read | xAI models and pricing |
What is the editor's rating for Grok?
7.5/10 overall. The reason for each score is in the table.
| Dimension | Rating | Why |
|---|---|---|
| Capability | 6.5/10 | The docs list text, voice, image, video and coding models with 256k to 1M token context 12, but no independent quality test was found. |
| Reliability | 4.5/10 | API rate limits by spend tier are documented 4, but no uptime figure, SLA or consumer usage-limit table was found. |
| Workflow fit | 7/10 | Web, iOS and Android apps with memory, Canvas and sharing are claimed by the vendor 5, and the API adds Batch, structured outputs and web search 2. |
| Pricing and value | 5.5/10 | API prices are published per model 1, but xAI's consumer plan prices could not be read and are reported only by third parties 8. |
| Privacy and security | 4.5/10 | The API docs state SOC 2 Type 2, no training on API data and 30-day retention 3, but consumer terms were not read and no report was seen. |
| Support | 4/10 | Rated from documentation only: developer docs exist 2, but no consumer support channel, response times or SLA were found. |
What test results have others published for Grok?
We found no independent test results for Grok that we could read and cite as of 2026-10-27. That is not evidence that it performs well or badly. Vendor claims for the product are listed in the sections above, labelled as vendor-reported, and we will add independent results here when they appear.
What security and compliance records exist for Grok?
xAI's consumer privacy policy and terms returned 403 to our fetch tool, so consumer training and retention terms were not read. The API security page makes no GDPR or ISO statement and does not claim HIPAA; it points to a BAA questionnaire.
Which certifications does the vendor list?
| Certification | Scope | Evidence | Verified |
|---|---|---|---|
| SOC 2 Type 2 | xAI API (vendor statement) | docs.x.ai (checked 2026-10-27) | No |
How is data handled?
| Question | Answer | Source |
|---|---|---|
| Is API data used to train models? | The vendor says it never trains on API inputs or outputs without explicit permission | docs.x.ai (checked 2026-10-27) |
| How long is API data kept? | 30 days by default for abuse auditing, then deleted, per the vendor; enterprise Zero Data Retention is optional and not recommended for most customers | docs.x.ai (checked 2026-10-27) |
| Where is API data processed? | The default endpoint does not guarantee a region; a US regional endpoint is documented | docs.x.ai (checked 2026-10-27) |
| Are consumer chats used to train models? | Not read from xAI. Third-party guides describe an opt-out in Data Controls; unverified | Not found on a page we could read |
What should you read next?
- Claude review: the closest general assistant in our queue.
- Perplexity review: the search-first alternative.
- Productivity: category criteria and the tools in the queue.
What changed in this review?
| Date | Change | By |
|---|---|---|
| 2026-10-27 | First version: API pricing, security FAQ and rate limits read from xAI docs; consumer pages, privacy policy, terms and plan prices could not be read and are marked unverified; test plan and evidence labels. | Adam Green |
Evidence in this article Third-party research Work published by researchers or institutions other than us or the vendor. We link it; we have not reproduced it.
Frequently asked questions
Has BestSuperintelligent tested Grok?
No. No hands-on test has been run. This review rests on xAI's public developer documentation, the xAI marketing copy supplied to us, and third-party pricing reports. Every feature listed is vendor-reported.
How much does SuperGrok cost?
We could not confirm it. xAI's consumer pages did not return prices to our fetch tool. Third-party sites report SuperGrok Lite at $10, SuperGrok at $30 and Heavy at $300 a month, but those are unverified. Check grok.com before buying.
Does Grok train on my conversations?
For the API, xAI's docs say it never trains on API inputs or outputs without explicit permission. For the consumer app and X, we could not read xAI's privacy policy; third-party guides describe an opt-out in Data Controls. Unverified.
Can Grok search the live web and X?
xAI's developer docs describe a Web Search tool that lets Grok search the web in real time and browse pages. The consumer description of X search with citations comes from xAI marketing copy and is vendor-reported.
Is Grok SOC 2 certified?
xAI's API security FAQ says it is SOC 2 Type 2 compliant and points NDA customers to a Trust Center. We did not see the report, and the statement covers the API, not the consumer apps. Unverified.
How much does the Grok API cost?
xAI lists per-million-token prices by model, for example grok-4.3 at $1.25 input and $2.50 output under 200,000 prompt tokens, and grok-4.7 at $2.00 and $6.00. Prompts at or above 200,000 tokens cost more. Checked 2026-10-27.
How is Grok rated?
Grok is rated 7.5/10. Its strongest area is workflow fit (7/10) and its weakest is support (4/10).
What is the weakest point of Grok?
Support rates lowest at 4/10. Rated from documentation only: developer docs exist, but no consumer support channel, response times or SLA were found.
Who is Grok for?
Individuals who want one assistant for chat, search and media generation, and developers who want an API with long context windows and published prices. Editorial interpretation. Look elsewhere if you need confirmed consumer data-use terms before pasting in work material, want a desktop-first writing assistant (see Claude), or want search answers built around sources (see Perplexity). Editorial interpretation.
Sources
- xAI docs: Models and pricing docs.x.ai, accessed .
- xAI docs: API overview docs.x.ai, accessed .
- xAI docs: Security FAQ docs.x.ai, accessed . The page names the company SpaceXSI in one statement; we have not confirmed how that relates to xAI.
- xAI docs: Consumption and rate limits docs.x.ai, accessed .
- Grok home page and xAI Grok page grok.com, x.ai, accessed . x.ai/grok returned 403 to our fetch tool and grok.com showed only interface text and a legal notice. The feature list in this review comes from xAI marketing copy supplied by the owner; it is vendor-reported and was not checked against a page we could read.
- X Help: About Grok help.x.com, accessed . Returned 403 to our fetch tool; read only through a search result summary. Unverified.
- xAI docs: Web search tool docs.x.ai, accessed . Tool pricing was not on the page we read.
- Grok pricing 2026 (third-party summary) ai-toolbox.co, accessed . Third party, read only through a search summary; one of several aggregators that agree on SuperGrok Lite $10, SuperGrok $30 and Heavy $300 a month. Not xAI. Unverified.
- BestSuperintelligent Claude review bestsuperintelligent.com, accessed . Internal page.
Related
- Same category: Claude review
- Same category: Notion AI review
- Same category: ChatGPT review
- Same category: Gemini review
- Same category: Grammarly review
- Same category: Microsoft Copilot review
- Category: Writing & productivity
- Where it sits in the top 50 AI coding tools by usage
- Every tool and model by editor's rating
- The full evaluation queue