Users report having their paid accounts suspended without clear reasons, warning, or way to appeal. The platform provides no direct communication channel and users find it nearly impossible to get their accounts reinstated. Multiple Trustpilot reviews describe being banned mid-subscription with no refund offered.

Support is handled through Discord channels and online docs, with no live chat, phone number, or real ticketing system. Reviewers describe support as unresponsive with automated replies. When issues arise with billing or bans, users have no effective way to reach a human.

Midjourney takes money, delivers defective service, and refuses refunds because 'the amount is too small to fight.' Users who get banned or cancel early report being unable to get any portion of their subscription refunded. Trustpilot rating sits at 1.6/5 stars largely due to billing complaints.

Your prompts and images are visible to other users in Discord by default. To keep work private, you must pay for Pro ($60/mo) or Mega ($120/mo) plans that include Stealth Mode. Basic and Standard users have no privacy option regardless of the nature of their work.

Companies making more than $1,000,000 USD in gross revenue per year must purchase Pro ($60/mo) or Mega ($120/mo) plans. This mandatory upsell catches businesses off guard when they scale and realize their Standard plan is no longer compliant.

Despite launching a web app in 2025, most features still route through Discord's 'hacky-ass interface' as Hacker News users describe it. Professional users find the Discord workflow unprofessional and difficult to integrate into real production workflows. Prompts and images are public unless you pay extra for Stealth Mode.

Midjourney doesn't accurately follow detailed prompts, especially when trying to achieve very specific outcomes. The system interprets prompts creatively rather than literally, which frustrates users who need precise results. Comic artists report getting 3-4 consistent character shots before the AI starts drifting.

Midjourney struggles with text generation - text in images doesn't achieve the same level of three-dimensionality as competitors like DALL-E 3. For projects requiring signage, product labels, posters, or any typography, DALL-E remains the default choice over Midjourney.

The prompt filter is described as overly sensitive and difficult to navigate. Terms like 'woman' or certain poses can trigger filters unpredictably, causing constant errors and blocking legitimate creative work. Users find the censorship rules inconsistent and frustrating.

Once an image is generated, you can't tweak specific parts without regenerating the whole thing. This lack of editing control wastes time and credits as users must regenerate entire images for minor adjustments. Competitors like Leonardo.ai offer canvas editors for precise control.

Despite improvements, Midjourney outputs can still be glitchy with generic art that lacks the uniqueness users seek. Multiple users report that 'evergreen prompts' are just fluke discoveries found through persistence and dumb luck, making consistent quality elusive.

Midjourney V7 still struggles with anatomical details, particularly hands, teeth, and background faces in crowded scenes. Users waste time regenerating images due to glitchy outputs. Even with V7's 30-40% improvement in bad generations, anatomy errors remain a consistent complaint.

Users get approximately 500-700 decent V7 images per month before hitting slow queues. Wait times can stretch to 5-40+ minutes during peak usage. Basic plan users especially feel the crunch with only 3.3 hours of fast GPU time monthly, which generates roughly 200 images.

Some users report that Midjourney V7 is somehow worse than V6.1 for their use cases. Fine artists and designers find the new version frustrating, with one user stating they're 'relieved that V7 is terrible' because it gives them an excuse to stop using the platform entirely.

Ethical issues related to ongoing copyright lawsuits create uncertainty for businesses using Midjourney commercially. The training data controversy and legal battles make some enterprise users hesitant to rely on the platform for commercial projects.