Midjourney

Basic Tier Ratings

Last Reviewed: 2026-07-14

Good image quality. ALL images PUBLIC. No privacy for marketing materials before launch.

Can create visual assets. ALL images PUBLIC - donors may see unreleased materials.

Can create illustrations. ALL images PUBLIC. ~3.3 GPU hours may run out.

Good for training visuals. Public images not usually a problem for training materials.

Excellent for social media visuals. Public images are fine for social content.

Can create email graphics. ALL images PUBLIC - competitors could see your campaigns early.

Good for event graphics and promotional materials. Public visibility usually fine.

Used for training

Per Privacy Policy (last updated June 2, 2025, unchanged as of July 2026), data collected through the Services is used in the process of training Midjourney machine learning algorithms. The TOS (effective June 23, 2025, unchanged as of July 2026) grants Midjourney a perpetual, irrevocable license to reproduce and distribute all Content you input and Assets you produce. As of June 11, 2026, V8.1 is the default model that all new generations are trained/served on; the same no-opt-out training policy applies. Stanford CRFM December 2025 report found zero transparency disclosures about training data practices.

No

No opt-out from AI model training. CCPA users can request data deletion at midjourney.com/account and opt out of data sharing with advertising partners. Deletion of account data does not necessarily remove data already used in model training.

Possible

Moderation team may review flagged or reported content. Community moderators (some volunteers) have access per public disclosures. No explicit policy on proactive human review of content.

Indefinite

Privacy Policy (June 2025, unchanged as of July 2026) states data retained as long as necessary for stated purposes and legal obligations. Users can request deletion at midjourney.com/account. Perpetual license granted to Midjourney for all content and assets survives account deletion.

No certifications

No SOC 2, HIPAA, ISO 27001, or FedRAMP certifications. RiscLens (January 2026) found no confirmed SOC 2 report. ALL IMAGES PUBLIC by default. Not listed on FedRAMP marketplace as of July 2026. Common Sense Media gave 0% for Data Security (evaluation unchanged since August 2023).

No Nonprofit Discount

Midjourney does not offer discounts or free plans for nonprofits or educational institutions. It is not listed on TechSoup or Goodstack. Consider Canva (free for 50 nonprofit users) or Adobe Express (discounted for nonprofits via TechSoup) instead.

All Images Public by Default

Critical privacy concern: Every image and video you create is visible in Midjourney's public gallery unless you use Stealth Mode. Others can see your prompts and outputs. Only Pro/Mega plans offer Stealth Mode ($60+/month).

No Free Tier

Free trial was removed in March 2023. You must pay to use Midjourney at all. Compare to Canva (FREE for nonprofits) or DALL-E (free tier available through ChatGPT).

Video Generation Now Available

Midjourney launched image-to-video generation in June 2025. Available on Standard, Pro, and Mega plans. Unlimited video Relax Mode requires Pro or Mega ($60+/month). Useful for social media reels, event promotions, and animated training content.

V8.1 Is Now the Default Model

As of June 11, 2026, V8.1 replaced V7 as Midjourney's default model at no extra subscription cost. It renders roughly 4-5x faster, offers cheaper/faster HD mode (native higher-resolution output), improved text rendering, and better prompt adherence, while remaining backward-compatible with V7 personalization profiles, moodboards, and style references.

Image/Video Generator, Not Writing Tool

Midjourney creates images and short videos from text prompts. It does NOT write grants, create documents, or help with text-based tasks. Pair with Claude for writing.

No Security Certifications

No SOC 2, HIPAA, ISO 27001, or FedRAMP certifications as of July 2026. Not listed on FedRAMP marketplace. Common Sense Media gave 0% for Data Security. Not suitable for healthcare or regulated environments.

Training Data — No Opt-Out

Midjourney's TOS grants a perpetual, irrevocable license to use all content you input and all images/videos you generate for any purpose including model training. Stealth Mode does NOT prevent training use. There is no opt-out mechanism on any tier.

New Hardware/Medical Division (Not Relevant to Core Tool)

In June 2026, Midjourney announced 'Midjourney Medical,' a new division building an ultrasonic full-body scanner and a planned wellness 'Midjourney Spa' (targeted for late 2027 in San Francisco). This is a separate consumer hardware venture unrelated to the AI image/video generation service and does not affect the pricing, privacy policy, or ratings covered in this review.

2026-06-17

Midjourney announced its first hardware product and a new division, "Midjourney Medical," building a full-body ultrasonic imaging scanner. A consumer wellness facility ("Midjourney Spa") is planned for Union Square, San Francisco, by late 2027. Unrelated to the core AI image/video generation product.

2026-06-11

V8.1 became the default Midjourney model on midjourney.com and Discord, replacing V7. V8.1 renders images roughly 4-5x faster (about 4 seconds in standard mode, 12 seconds in HD), supports HD mode at twice the size/4x resolution of V7 at a much lower relative cost than the original V8 Alpha, and maintains compatibility with V7 personalization, style references, and moodboards.

2026-04-14

V8.1 Alpha released on alpha.midjourney.com, improving on V8.0 with a more V7-like aesthetic, more stable moodboards/style references, cheaper and faster HD mode, restored image prompts, and a new prompt shortener.

2026-04

V8.0 Alpha model deprecated approximately two weeks after V8.1 became available, per Midjourney's release notes.

2026-03-17

V8 Alpha launched on alpha.midjourney.com. Features: ~5x faster generation, native 2K resolution (--hd mode), improved prompt adherence, improved text rendering. Not yet available on main site or Discord at time of launch.

2026-02

Rooms page removed from Midjourney website. Social and collaboration features planned for future replacement.

2026-01-09

Niji V7 launched — Midjourney's anime/Asia-tuned model. Features improved coherence, prompt understanding, text rendering, and style reference performance.

2025-06-23

Terms of Service updated (effective date June 23, 2025; unchanged as of July 2026). Formalized commercial use licensing, explicitly codified the $1M revenue threshold in main TOS, and updated IP disclaimers in light of ongoing AI copyright litigation.

2025-06-19

Video generation (image-to-video) launched. Available on Standard, Pro, and Mega plans. Unlimited video Relax Mode only on Pro and Mega. Videos capped at 720p HD, up to 21 seconds.

2025-06-17

V7 became the default Midjourney model. V7 introduced Draft Mode, Omni Reference, improved image quality, and better body/hand coherence.

2025-06-02

Privacy Policy updated (unchanged as of July 2026). Explicitly acknowledges data collection through ML training process. No opt-out from training data use.

For organizations with fewer than 5 staff

Not recommended. Canva is FREE for nonprofits and includes AI image generation tools. Midjourney has no nonprofit program, no free tier, all images are public by default, and requires $60/month for privacy via Stealth Mode. Standard plan ($30/month) includes video generation but still no image privacy. The June 2026 switch to V8.1 as the default model improves speed and quality but does not change any of these fundamental limitations.

For organizations with 5+ staff

If budget allows and high artistic quality or video generation is essential, Pro plan ($60/month) provides Stealth Mode for privacy plus unlimited image and video Relax Mode, now powered by the faster V8.1 model (default since June 2026). But Canva FREE is usually sufficient for most nonprofit visual needs. Factor in that Midjourney has no security certifications and grants itself a perpetual training license on all content.

For health data or vulnerable populations

Do not use. No HIPAA compliance, no BAA, no SOC 2 certification. All images are public by default (requires paid Stealth Mode). Midjourney retains a perpetual license to use all generated content. Not suitable for any use case involving protected health information or sensitive client data. Note: Midjourney's newly announced 'Midjourney Medical' hardware division (ultrasonic body scanner) is an entirely separate consumer product and does not change this guidance for the AI image/video generation service.