Midjourney review

admin
Midjourney review cover graphic

Midjourney is one of the best-known AI image generators. You describe an image in text, and it returns a set of candidate pictures that you can vary, upscale, edit or turn into short video clips. It has grown from a Discord-only service into a product with a web Create page, and its current default model is called V8.1.

This review covers what Midjourney’s own documentation says about its features, plans, privacy settings and training data, plus one reported legal dispute that buyers should know about. It is written for people deciding whether to pay for it, not for people who already use it every day.

How we researched this: we did not run hands-on tests for this article. It is based on the vendor’s own documentation and public sources, checked on 11 November 2026. Features and prices change, so confirm details on the vendor’s site before you buy.

Key takeaways

  • Midjourney sells subscriptions measured mostly in “Fast” GPU time rather than a fixed number of images, and unused Fast time does not roll over.
  • Four plans exist (Basic, Standard, Pro, Mega). Unlimited “Relax” image generation starts at Standard, and Stealth Mode (private creations) needs Pro or Mega.
  • Creations are public by default and can appear on the Explore page, so privacy is a paid feature, not a default.
  • Midjourney says you own what you create, and that companies earning more than 1,000,000 US dollars a year must use Pro or Mega.
  • A copyright lawsuit from film studios was still ongoing as of July 2026 reporting, and Midjourney’s published training summary says user prompts and interactions were used for training.

What Midjourney is

Midjourney is a text-to-image service run by Midjourney, Inc. Its older getting-started documentation describes a Discord workflow: you create a Discord account, subscribe on midjourney.com, join the Midjourney server and type the /imagine command followed by a prompt. You then receive a grid of four options and use buttons to upscale, vary, zoom out or pan.

The current documentation also describes a web Create page on midjourney.com. It has an Imagine bar for prompts, an image upload menu (10MB maximum per image), a video or still toggle once an image is in your prompt, a settings menu for aspect ratio, Standard or HD model, Raw mode and style controls, and a live creation feed. Discord is still supported, but the web page is now the main hub described in the help centre.

The current model

Midjourney’s update post says V8.1 is the new default image model, replacing V7. It claims more coherent images, closer following of detailed prompts and better text rendering, and says HD mode gives images twice the size and four times the resolution of V7 output. The post does not give a date, and the speed figures it quotes (about 4 seconds in standard mode and 12 seconds in HD) are the vendor’s own claims, so treat them as unverified. The post also says V7 omni-reference stays available while an improved V8 version is being trained.

Who it is for

Based on the documented features, Midjourney suits people who want to make stylised, visually rich images from text and are comfortable iterating: running a prompt, choosing a result, varying it and trying again. That includes designers looking for concept art and mood boards, marketers who need illustrative assets, writers who want cover or scene ideas, and hobbyists.

It is less obviously suited to teams that need strict control over a precise layout, guaranteed privacy by default, or a vendor that offers legal indemnity. If you are new to prompting, our beginner’s guide to AI image prompts explains the basics that apply to any generator.

Key features (documented)

The features below are those described in Midjourney’s help centre and documentation pages. We have not verified how well any of them work.

Feature What the documentation says
Text to image Prompts produce a grid of four image options. Upscale and variation tools work on a chosen result.
Zoom out and pan Extends the canvas of an image without changing the original content.
Image blending and uploads The /blend command mixes uploaded images. The web page accepts uploads up to 10MB.
Editor Plans include use of the Editor on uploaded images.
Video An image can be turned into video. Basic includes SD video; HD video starts at Standard.
Personalisation and references Help centre sections cover Personalization, Omni Reference, Draft and Conversational modes, and an Edit model.
Speed modes Fast, Relax and Turbo, described below.

Fast, Relax and Turbo

Midjourney’s GPU speed page explains the three modes. Fast mode uses a monthly allotment of GPU time that resets at each renewal with no rollover; a typical prompt costs about one minute, with upscales and unusual aspect ratios costing more. Relax mode uses unlimited GPU time instead of Fast time, is available on Standard, Pro and Mega, and queues requests that can wait from 0 to 10 minutes. Turbo draws on a high-speed pool that can be up to four times faster than Fast, but uses Fast time at double the rate. The page calls Turbo experimental, and the plan information pages say Turbo is not currently supported in V8.1.

Pricing model

Checked on 11 November 2026. All prices are in US dollars. Plans renew automatically unless you cancel, and yearly billing is listed as a 20 percent discount. Confirm current figures on Midjourney’s own pricing page before you buy.

Plan Monthly Yearly (total) Fast time Relax Stealth Mode
Basic $10 $96 3.3 hours per the plans documentation Not listed No
Standard $30 $288 15 hours Unlimited images No
Pro $60 $576 30 hours Unlimited images and SD video Yes
Mega $120 $1,152 60 hours Unlimited images and SD video Yes

The two official pages describe Basic slightly differently. The pricing page describes it as limited to about 200 image jobs a month (each job producing four images), while the plans documentation lists 3.3 Fast hours. These may be two ways of describing the same allowance, but we could not confirm that, so check the live page.

What else affects cost

  • Extra Fast time costs $4 an hour on every plan, according to the plans documentation.
  • Unused Fast time does not carry over to the next month, on monthly or yearly plans.
  • HD video is available from Standard upward, and unlimited SD video in Relax mode needs Pro or Mega.
  • Simultaneous Fast jobs rise by plan, for example 3 image jobs on Basic and Standard and 12 on Pro and Mega.
  • The documentation does not mention a refund policy on the plans page and points to the Terms of Service for that.

Our guide to reading an AI tool’s pricing page explains why GPU-time pricing is harder to compare than per-image or per-seat pricing.

Trials

Midjourney’s help centre says no free trial is currently offered on Discord, and that a limited trial is available in its niji journey app on iOS and Android. We could not confirm the exact terms of that trial.

Strengths

  • Clear tiering. Four plans with a published ladder of Fast hours and concurrent jobs make it easy to see what each step buys.
  • Unlimited Relax mode. From Standard, you can keep generating images without spending Fast time, if you can accept queue waits of up to 10 minutes.
  • Commercial use is addressed. The help centre says you own the images and videos you create, even after cancelling, and the plans page says you can use them in just about any way you want if you have subscribed at any point.
  • Web and Discord options. You can work from the web Create page or from Discord.
  • Image and video in one place. Documentation covers both still images and video from the same account.
  • Transparency on training data. Midjourney publishes a public summary of training content, which not every vendor does.

Limits and cautions

  • Fast time expires. Heavy users can run out before renewal, and light users lose what they do not spend.
  • Privacy costs more. Stealth Mode is limited to Pro and Mega.
  • Content rules apply. The older quick-start page says content rules apply across all tools, including a ban on gore and adult content.
  • Pricing is not easy to predict. Because the unit is GPU time, the number of images you get depends on settings, upscales and aspect ratios.
  • Legal uncertainty. See the section on the studio lawsuit below.
  • Some details conflict between pages. We found the Basic allowance described two ways, and one documentation page we tried to read (the plan comparison article) did not load, so some detail comes from other official pages.

Privacy and data

Public by default

Midjourney’s help centre says the service is open by default: any image or video you generate may appear on the Explore page at midjourney.com, including work made in private Discord servers and in direct messages with the bot. Stealth Mode hides your creations from other users on the website, requires Pro or Mega, and still requires your content to follow the Community Guidelines. It covers midjourney.com only, so work created in a shared Discord channel can still be seen by others in that channel.

If you plan to upload client material or unreleased product images, read our guide to protecting company data when using AI tools first.

Training data

Midjourney’s public summary of training content (version 1, updated 17 March 2026) lists several sources: publicly available datasets, licensed data, other private third-party data, crawled data collected from 2023 onward, synthetic data and user data. It states that user prompts and interactions were used for training under its Privacy Policy and Terms of Service. It says crawlers respect robots.txt and that the model is continuously trained. It also says Midjourney is not a signatory to the EU GPAI Code of Practice. Read the summary yourself if training data matters to your organisation.

The studio lawsuit

This is general information, not legal advice. The Art Newspaper reported on 9 July 2026 that a copyright suit brought in 2025 by Disney, Universal and Warner Bros against Midjourney was ongoing in a California federal court. According to that report, the studios allege Midjourney trained on images that let users generate copyrighted characters. Midjourney’s defence is fair use, and in June 2026 the court denied its request for discovery into the studios’ own AI use; on 29 June 2026 Midjourney asked for that ruling to be reviewed. We did not find a final outcome, and what it would mean for customers is unclear.

Alternatives

We did not research alternatives for this article in enough depth to rank them, so these are starting points, not recommendations. Another review in this series covers Adobe Firefly, which publishes a different approach to training data and credits. You can also use a general image feature inside an AI assistant you already pay for. Whatever you shortlist, use our guide to comparing AI tools side by side fairly so each one gets the same prompts and scoring.

Who should and should not use it

Best for Look elsewhere if
Individuals and small teams exploring visual ideas who are happy to iterate You need work to be private by default and cannot justify Pro or Mega
People who want one subscription for images and short video Your legal team requires indemnification or fully licensed training data
Users who value Relax mode for high-volume, low-urgency generation You need predictable per-image costs
Businesses under the 1,000,000 dollar revenue line who accept the terms Your company is above that line and cannot buy Pro or Mega

Questions to ask before buying

  • How many images or clips will we make each month, and will Fast hours cover them or will we live in Relax mode?
  • Do we need Stealth Mode, and is that worth the jump to Pro?
  • Is anything we upload confidential or covered by a client contract?
  • Does our legal team accept the published training summary and the status of the studio lawsuit?
  • Do we meet the revenue threshold that forces Pro or Mega?

Before paying, run a short trial of the tool on your real use case, and use our guide to choosing an AI tool to set your criteria first.

Frequently asked questions

Is Midjourney free?

We found no standard free plan. The help centre says there is no free trial on Discord and a limited trial in the niji journey app. Check the current pricing page for changes.

Do I own the images I make?

Midjourney’s help centre says you own the images and videos you create, even after cancelling. Companies with over 1,000,000 US dollars in gross annual revenue must use Pro or Mega. Read the Terms of Service for the full wording.

Can other people see my images?

By default, yes: creations may appear on the Explore page. Stealth Mode on Pro or Mega hides them on the website.

What happens to unused Fast hours?

They expire at the end of the billing cycle, including on yearly plans.

Does Midjourney use my prompts for training?

Its public training summary says user prompts and interactions were used for training. We did not find an opt-out in the pages we read, so ask the vendor if this matters to you.

Can I rely on AI images being accurate?

Generated images can contain errors, such as distorted details or text. See our guide to spotting AI-generated text and images, and review every image before publishing.

Conclusion

Based on its documentation, Midjourney is a mature, well-documented image and video generator with a straightforward plan ladder, a commercial-use statement and a published training summary. Its main trade-offs are public-by-default creations, expiring GPU time and an unresolved copyright dispute. It is best for people who iterate on visuals and accept those terms.

If you are weighing it against other tools, start with our guide to choosing an AI tool and the side-by-side comparison method, then confirm every price on the vendor’s page.

Sources

Keep reading