Midjourney V8.2: AI Creatives for Ads in 2026

Midjourney leveled up in 2026, and many are still operating with the 2024 playbook. Since July 24, the default model is V8.2, the web app made Discord optional, and draft mode slashed the cost of exploring ideas. In this guide, I'll walk you through what's changed since June, current pricing, where the real legal risk lies, and how we fit the tool into our clients' ad creative production.

TL;DR

• V8.2 is Midjourney's default model since July 24, 2026 — focus on aesthetics, less disposable generation, and more precise personalization.
• June was the heavy month: V8.1 became default on 06/11, draft mode arrived on 06/16, and --sref random on 06/25.
• Midjourney plans range from $10 to $120/month (20% off annual), with 3.3 to 60 hours of fast GPU and overtime at $4.
• All paid plans include commercial rights, but companies with gross revenue above $1 million/year need Pro or Mega.
• The Disney, Universal, and Warner lawsuit is still in discovery, with no trial date set. The practical risk isn't using the tool — it's generating someone else's characters.

What changed in Midjourney between June and August 2026

If you stopped following Midjourney after V7, you missed three updates that change the workflow, not just image quality. The timeline, according to the official release notes, went like this:

1. June 11, 2026 — V8.1 becomes the default model. Renders at double the size and four times the resolution of V7, with about 4 seconds per image in SD and 12 seconds in HD.
2. June 16 — draft mode. Generates 24 low-resolution images at once, with a "Vary" button to promote only the chosen ones to full quality. Also debuted the --preview flag, which unlocks test features before launch.
3. June 25 — --sref random. Applies 24 random styles to the same prompt, turning art direction exploration into a two-minute task.
4. July 24 — V8.2 becomes the default. The team itself described the goal as more creative, bold, and sophisticated images, with a sharp drop in bad results and personalization that better understands each account's taste.

V8.1 remains selectable, which matters more than it seems: if you calibrated moodboards and style references on it, the automatic switch to V8.2 could shift your results. In August, the pace cooled — the August 5 office hours passed without a product announcement, which usually signals consolidation, not stagnation.

Why Midjourney's draft mode matters more than the model jump

Before, testing an art direction cost fast GPU time on every attempt. With draft mode, you burn almost nothing to see 24 paths and only pay full price for the two or three images that make it into the ad. In practice, a Midjourney exploration batch that used to take half an hour of fast time now fits in minutes. For teams running weekly creative tests, this reduces the cost per tested variation — the metric that really matters in paid media.

How much does Midjourney cost in 2026 and which plan makes sense

Midjourney's pricing structure remains subscription-based, with payment in dollars and international card billing — factor in any foreign transaction fees when approving the budget. The annual plan is 20% cheaper.

Plan	Monthly	Annual (per month)	Fast GPU Hours	Unlimited Relax	Video
Basic	$10	$8	3.3 h	No	SD only
Standard	$30	$24	15 h	Image	SD
Pro	$60	$48	30 h	Image and SD video	SD and HD
Mega	$120	$96	60 h	Image and SD video	SD and HD

Some useful math before you choose. A static image on Midjourney consumes roughly one minute of GPU; an HD video batch consumes about 26 minutes. Overtime fast costs $4 on any plan. In other words: on Basic, three HD video batches already blow through the entire monthly allowance.

For most marketing operations, Standard at $30 is the sweet spot — the 15 fast hours cover what's on deadline, and Relax mode absorbs the rest at no extra cost. Only move up to Pro when video becomes routine or when the revenue rule (next section) requires it. Mega is for studios with high-volume video production.

Commercial rights: the license allows it, but there's a $1 million cap

This is the part I see most companies ignoring when adopting Midjourney. All paid Midjourney plans grant commercial usage rights to generated images — you can sell them, use them in campaigns, deliver them to clients. Free trials do not grant this right.

The costly detail is in the size: companies with annual gross revenue above $1 million must subscribe to Pro or Mega for commercial use. Using Basic or Standard in that range violates the terms. Roughly, that's a threshold that catches many agencies and mid-sized e-commerce operations — and the difference between $30 and $60 a month will never justify a contractual dispute with a client.

Two practical recommendations we apply here:

• Subscribe under the legal entity that actually uses it, not the designer's personal card. If a client audit asks for proof of license, you need an invoice in the right name.
• Log the prompt and style parameter for each published piece. This serves both to reproduce the art and to demonstrate authorship of the creative process.

This isn't legal advice: terms change and the official text is what counts. But ignoring the $1 million threshold is an unnecessary risk for $30 a month.

Disney, Universal, and Warner vs. Midjourney in the US: the risk lies in the prompt

In June 2025, Disney, NBCUniversal, and DreamWorks filed a copyright infringement lawsuit against Midjourney, alleging the platform trained models on protected works and allows users to generate characters like Darth Vader, Elsa, Bart Simpson, and the Minions. On November 4, 2025, the court consolidated the case with the lawsuit filed by Warner Bros., with Disney's case designated as the lead.

The current stage, as of August 2026, is discovery — the evidence exchange phase. A judge ruled that studios only need to provide information about AI use related to consumer-facing content, a decision the defense appealed, arguing it restricts their counter-evidence material. There is no merits ruling or trial date.

Translated for ad producers: the risk isn't using Midjourney in your daily agency work. The risk is asking for a third party's character, mascot, or visual identity in the prompt and publishing that in paid media. That piece is indefensible regardless of the tool that generated it. Block any nominal references to franchises, celebrities, and competitors in your internal process — including disguised ones like "in the style of."

FTC and regulation in the US: who answers for AI-generated ads

Here the landscape is more concrete than the American litigation. The FTC has been active in AI-related advertising. The FTC's guidance on AI and advertising makes clear that existing consumer protection laws apply to AI-generated content. The FTC Act prohibits deceptive practices, and the FTC has issued warnings about AI-generated deepfakes and misleading endorsements. Additionally, state laws like CCPA and others may impose obligations. The key points:

• Joint liability. Advertisers, agencies, and influencers are jointly responsible for the truthfulness and compliance of AI-generated or AI-edited content.
• No misleading product attributes. A generated image cannot suggest texture, size, result, or composition that the real product doesn't deliver.
• Labeling is not yet mandatory, but the FTC encourages transparency about synthetic content, especially when the audience includes children or teens.

On the legislative front, the American Data Privacy and Protection Act (ADPPA) has been proposed but not yet passed. Meanwhile, the FTC continues to enforce against deceptive AI practices. Setting up a creative approval process now, with prompt logging and documented human review, will put you ahead when regulations tighten.

Midjourney Medical: why the company entered ultrasound

On June 18, 2026, in San Francisco, the company announced a health division and an imaging device called Fullbody Ultrasonic Computational Tomography, nicknamed "Ultrasonic CT." The proposal is to scan the entire body in about 60 seconds using sound waves and water, without radiation or magnetic fields. The announced investment exceeds $74 million.

The numbers from the project warrant caution: the current prototype takes about 20 minutes per exam, has been used on roughly a dozen people, has no FDA clearance, and the first unit in San Francisco is expected by the end of 2027.

Why does this matter to someone who just wants ad images? Because it says something about risk concentration. The Midjourney team is spreading capital and attention outside image generation while facing a billion-dollar lawsuit. It's not a reason for panic, but rather not to depend on a single vendor. Keeping an alternative pipeline running — like a ComfyUI pipeline on Google Colab — costs little and ensures that a change in price, policy, or availability doesn't halt creative production on the eve of a campaign.

How we fit Midjourney into the paid traffic creative workflow

In 15 years developing for clients in the US and abroad, I've learned that a good tool without a process becomes rework. The workflow we use today, in six steps:

1. Fix the identity before the prompt. Build a moodboard with 5-10 approved brand pieces and extract the style reference from there. That's what prevents each round from looking like a different company.
2. Explore in draft mode. Run the concept with --sref random and look at the 24 thumbnails as a block, not one by one. Art direction is decided by comparison.
3. Promote only three. Bring to full resolution only the options that survived the side-by-side comparison.
4. Adjust in the web editor. Composition, cropping, and additional elements are handled in Midjourney's editor without leaving the browser — Discord is now optional.
5. Export in the formats the auction consumes. 4:5 for feed, 9:16 for stories and reels, 1:1 for catalog. Generating in 16:9 and cropping later destroys framing.
6. Document prompt, seed, and style reference in the same spreadsheet as the creative test. Without this, the winning creative isn't reproducible next month.

When not to use AI-generated images

• Real product photos. A client receiving something different from the ad opens a complaint, and liability is joint.
• Faces of people in testimonials. Synthetic social proof is an ethical, legal, and reputational problem all at once.
• Before-and-after in health, aesthetics, and finance. Heavily regulated territory.
• Software screenshots or numerical data. Models still get small text wrong and invent numbers.
• Any piece where truthfulness is the selling point.

Five common mistakes that ruin Midjourney creatives

The pitfalls I see most in account audits:

• Running without a fixed style reference. Result: 40 beautiful images and no brand consistency across the funnel.
• Asking for competitor or franchise aesthetics. Beyond the legal risk, the ad reinforces recall of the other brand.
• Ignoring small-screen readability. The creative is approved on a 27-inch monitor and consumed on a 6-inch screen, with a thumb covering a corner.
• Leaving text in the generated image. Even with Midjourney's V8.2 advances, typography is a graphic editor's job. Generate clean art and compose text on top.
• Testing one piece at a time. Without three to five simultaneous variations per set, there's no statistical learning — just opinion.

The same logic applies as we discussed in AI agents for businesses: AI accelerates execution but doesn't replace the judgment of those who understand the business. And, as I wrote in the analysis of Gemini news at I/O 2026, the competitive advantage lies in the process around the model, not the model itself.

Conclusion: what to do in the next 30 days

If you already use Midjourney in your operations, do three things this week: confirm your plan matches your company's revenue, redo your moodboards in V8.2 before the aesthetic difference shows up mid-campaign, and create the prompt, seed, and style reference spreadsheet. If you don't use it yet, start with the Standard plan, run a two-week creative test against your current image bank, and compare cost per result — not per image.

If you'd like to discuss structuring this production within your paid media operation, that's exactly the kind of project we handle at Agathas Web: process, automation, and measurement tied to what generates revenue, not what generates pretty screenshots.