What Is Midjourney? Model Family Map

Updated: 2026-08-29. Public product names, default versions, parameters, and capability bounds follow midjourney.com and in-product docs/announcements that day. Third-party entries are not Midjourney official.

Introduction

Midjourney is a standalone image (and extended video) generation product for creators—not a side button inside someone else’s chat model, and not a generic label for “any drawing site.” You can generate on the official Web Create page, use Discord bot commands when that path is still available for your account, or run learning drafts through a China-friendly third-party workbench whose risks you have already reviewed. The usual traps: treating a nickname on an entry button as “the official default weight that day,” or treating one pretty demo as shippable brand asset.

What this guide solves

• One-sentence mental model: product, model version, and entry layers
• How V8.2, V8.1, Niji 7, and Video divide labor
• Differences among Web / Discord / third-party entries—and why there is no official public API
• Capability bounds: hallucinated layouts, garbled text, copyright, and who owns responsibility
• A three-step selection method plus a reproducible sample set—not one demo reel

One-sentence definition

Midjourney = an independent text-to-image (and adjacent) product brand + multi-generation models / modes (such as V8.2, V8.1, Niji 7, Video) + multiple delivery entries (official Web, Discord, and third-party wrappers). Which models you can use, what defaults to, which switches exist, and how subscription/credits work all move with releases. 2026 comparisons here are for navigation; ship against the official list that day.

Product vs model: keep them separate

Layer	Examples	Correct use
Brand / product name	Midjourney	Internal talk, guide titles, requirements docs
Model / mode	V8.2, V8.1, Niji 7, Video	Selection, comparison tests, delivery specs with a version written down
UI options	Create-page dropdowns, Discord parameters	Follow current UI copy; screenshots age out
Third-party button labels	“Midjourney Vx” on some site	Marketing name for that entry—not an official weight guarantee

Common mistakes:

1. Treating a “Midjourney” tag on a third-party page as a specific official generation. Routing may differ—see the China access guide.
2. Copying an old default version from a blog into production specs. Defaults switch; re-check midjourney.com before release.
3. Assuming anything the web UI can do equals an official public API. Midjourney does not offer a publicly available official developer API (per statements that day).

Memory aid: say Midjourney V8.2 to people; write “entry + UI-shown version + date” for acceptance; bill and subscribe against the official site.

If a requirements doc says both “use Midjourney” and “version: …,” reviewers must be able to reconcile both lines. If they cannot, stop and verify on the official site—do not invent from memory. Third-party screenshot button copy alone is not enough for procurement or acceptance.

V8 line vs Niji vs Video (2026 navigation)

Name	Typical lean	Better for	Less suitable for	Dig deeper
V8.2	Flagship all-rounder; photo-real and design balance (per official default notes)	Everyday finals, product concepts, marketing heroes	Locking “pure anime line art” onto a photo-leaning default	V8.2 full guide
V8.1	Same-gen prior tier / comparison baseline	Aligning with existing assets; version regression	New projects that refuse upgrades forever with no written reason	This page + official version notes
Niji 7	Anime / comic-illustration aesthetics	Character sheets, manga-panel feel, Japanese illustration	Sole source for hard photo-real product photography	Niji 7
Video	Stills into short motion / video drafts (capability per site that day)	Motion previews, short-shot concepts	Treating it as a full editorial + finishing pipeline	Official Video feature notes that day

How to read the table:

• Product names show up in UI copy, guides, and community talk—useful for humans.
• Parameters and switches (stylize strength, personalization, aspect) may differ by version visibility; follow official docs that day.
• The same name may map late—or be temporarily unavailable—on different entries; trust what your account can see.
• Price, credits, queue speed: no hard-coded numbers; follow billing and subscription pages that day.

V8.2: common flagship default

When people ask “what is the official default right now,” 2026 conversations often point to V8.2 (still confirm on your Create page after login). It usually carries the “quality-first everyday workhorse” role: product-studio feel, environment concepts, portrait mood, design-led layouts. System mastery belongs in the dedicated guide—do not stop at noun level here.

V8.1: neighbor baseline, not “dead weight”

V8.1 is often used for same-prompt differences against the adjacent version, or to keep a team’s already-validated prompt habits alive. If your brand memory was trained on a batch of V8.1 frames, jumping to V8.2 can feel “overall better but style-shifted.” The fix is a small sample-set comparison—not a meeting argument about “newer = always better.”

Niji 7: dedicated anime track

Niji leans anime, illustration, and character design. Mixing tracks fails in predictable ways: forcing Niji into hard photo-real metal products, or forcing V8 into thick Japanese comic paint and complaining it “doesn’t feel anime.” Pick the aesthetic track first; tune parameters second.

Video: stills into motion drafts

Video-related capabilities let creators extend a single frame into a short motion preview. They solve early validation of motion direction and camera feel—not replacement for professional compositing and grade. Feature switches, duration caps, and billing follow official notes that day; these guides do not freeze aging numbers.

Before Video, start from a compositionally correct still: motion will not “fix” wrong subject proportions or warped trademarks—it only makes them more obvious on a timeline. Write shot intent clearly (push, pull, mild parallax, subject still / background moves). Final delivery still goes through edit, titles, and sound.

How to read version evolution (ignore marketing spin)

Each public generation usually trades off default aesthetics, detail obedience, text behavior, speed, and credits—not a simple “crushes the previous gen everywhere.” When you read release notes, grab three lines: did the default switch, which old parameters still work, does personalization change. Community screenshots of “must-on parameters” are usually bound to a specific personalization stack and subject—pasting them into brand specs is risky.

For teams, maintain a short version changelog: date, default version, sample-set score deltas, whether a full cutover is allowed. That beats arguing “V8.2 is always prettier than V8.1” in meetings.

Web, Discord, and access paths

Path	What it solves	Watch outs
Official Web	Authoritative UI for defaults, subscription, Create, personalization	Reach depends on network and account policy
Discord	Command-style workflow, community, historical habits	Command format and availability follow that day’s guidance
China third-party	Learning-draft reach	≠ official; model labels may lag (e.g. still showing V7)

• Official: www.midjourney.com
• China-friendly (third-party): Midjourney V7
• Mirror studio: Text-to-image studio

Steps, risks, and troubleshooting for China access: China access guide. Cluster navigation: Guides hub.

No official public API: what that means

As of this guide’s writing, Midjourney does not provide a publicly available official developer API. Therefore:

• Do not write “call Midjourney’s official REST / SDK” in a technical design and invent a docs link.
• If a product needs programmable batch generation, evaluate other image models with public APIs—or wait for Midjourney’s formal open access and read those terms.
• Third-party “Midjourney-style / Midjourney-named” interfaces contract with that platform, not Midjourney official support.
• Scraping the official UI with automation usually violates terms and is extremely brittle; these guides neither encourage nor document that path.

One line: Midjourney is a creation product, not a public cloud inference API SKU (per statements that day).

If a roadmap says “integrate Midjourney,” rewrite it into something executable in review—for example: “Designers generate on official V8.2, human-accept, then upload to CMS,” or “Evaluate other image models with public APIs and SLAs.” Vague “system auto-calls Midjourney” with no official interface docs will stall procurement and security review.

Capability bounds: “can generate” ≠ “can ship”

Image models produce pixels that look plausible. You may still see extra fingers, warped trademarks, unreadable pseudo-text, bad perspective, style drift from a reference, or invented UI controls. Build fixed habits:

• Text: if on-image copy is required, put the exact string in the prompt and proofread by hand; important titles are often safer in layout tools.
• Brand: for logos, mascots, packaging rules, supply licensed references and say “keep marks sharp and undistorted.”
• Likeness: avoid “generate [living public figure]” for disputed uses; commercial portraits need authorization and compliance review.
• Facts & dates: poster dates and regulatory slogans must come from the business owner of record—the model does not maintain truth.
• Privacy: IDs, undredacted client photos, unpublished design sources do not belong on untrusted entries.
• Publish responsibility: copyright, advertising law, platform rules, and final sign-off sit with people or orgs; the model accelerates drafts.

Task type	Prefer	How to verify
Social art, blog heroes, product concepts	V8.2 (or official default flagship that day)	Check composition and crop safe zones against use
Anime characters, illustration covers	Niji 7	Character feature checklist + style consistency
Version migration / regression	V8.1 ↔ V8.2 side-by-side	Fixed sample-set scores
Short motion preview	Video (if your account has it)	Motion direction matches storyboard intent

Three steps to choose a tier

1. Define task type and acceptance criteria
   Exploring composition, shipping a publishable final, or motion preview? Write what “good” means: aspect, whether clear on-image text is required, faces allowed, hard brand colors. Without acceptance criteria, every version feels like a lottery.

2. Rank hard constraints
   Aesthetic track (photo-real vs anime), speed, credit burn, China reach, whether data may leave the org—which cannot break? Users in China often put “can we open an entry at all” on that list. When constraints conflict, split stages: third-party drafts, official cross-check finals.

3. Build a small sample set before you compare
   Prep 8–20 real briefs, fix prompts and aspect, run V8.2 / V8.1 / (when needed) Niji on the same entry. Log obedience, text error rate, style drift, and human fix time. One pretty demo is not a portfolio.

Selection test prompt (copy-ready):

Use: [final placement]
Aspect: [e.g. 16:9]
Subject & scene: [full description]
Style & light: [concrete; avoid empty adjective piles]
On-image text (if any): "exact copy"
Hard bans: [watermarks / garbled glyphs / warped marks / …]
Obey constraints strictly; if information is missing, ask at most 3 clarifying questions before generating.

Text-rendering comparison template:

Generate a clean landscape event hero.
Main title text must be exactly: "Spring Member Day"
Subtitle: "Members only · limited window" (including the middle dots)
Style: clean flat-design feel, large color blocks, generous whitespace, crop-friendly for mobile.
Ban: typos, English gibberish, extra promo numbers.

When to pick which tier (quick decisions)

• You only know you need “good images” for marketing/content → start on V8.2 and run the acceptance checklist.
• Clear anime / character sheet → Niji 7, then refine prompts.
• Team already has heavy V8.1 assets and prompt habits → sample-set compare first, then decide on a full V8.2 cutover.
• Need short motion feel → check Video capability that day; still lock the still composition.
• Network in China is unstable; learning prompts first → third-party may help, but model labels can lag; see China access.

Completion checklist

• You can explain in one sentence: “Midjourney is the product; V8.2/Niji are models/modes”
• You recorded current entry domain + UI-shown version (with date)
• You know: no official public API—third-party interfaces are not Midjourney capability
• You prepared at least 5 real tasks for a sample set—not one demo
• You read the Guides hub and know the next deep guide
• Commercial use has a rights/compliance review planned; generated pixels are not legal advice

Quick access

• China-friendly (third-party): Midjourney V7
• Mirror / text-to-image studio: Text-to-image studio
• Official site: www.midjourney.com

Third-party ≠ official; whtai may still list V7 while the official default is V8.2—trust each list that day.

FAQ

Is Midjourney the same as ChatGPT image generation?

No. Different products, different model families, different subscriptions and terms. Treating “both can draw” as one workflow misaligns prompt habits and acceptance criteria.

Why does a third-party still show V7 while this guide talks about V8.2?

Release cadence and third-party routing updates often diverge. You can learn prompt structure on a reachable entry; confirm default flagship and formal delivery on the official site.

Can Niji 7 replace V8.2 for product photography?

Usually not as a sole source. Niji leans anime aesthetics; hard photo-real products and environment concepts prefer the V8 line, then human retouch.

Are parameters like —v and —stylize still valid?

Names, defaults, and availability change by version. Follow official docs that day—do not memorize an expired blog’s parameter table.

Who owns copyright in generated content?

Follow the Midjourney terms you accepted and the law where you operate; third-party entries add their own terms. These guides are not a substitute for counsel.

Official resources

• midjourney.com
• In-product docs, version notes, subscription, and community guidelines (whatever is visible after login that day)

Further reading

• Midjourney Guides Hub
• Midjourney China access guide
• Midjourney V8.2 full guide
• Midjourney Niji 7 selection & practice
• Midjourney getting started: first image
• Midjourney prompt playbook

Summary

Midjourney is a standalone image product; V8.2 / V8.1 / Niji 7 / Video are differently oriented models and modes—not interchangeable nicknames. Entry solves “can open,” model solves “how it looks,” personalization solves “does it feel like your taste”—keep the three apart. No official public API; third-party ≠ official. Select with acceptance criteria and a small sample set, not one “god prompt.” Defaults and parameters always follow the official site that day.