Midjourney is a text-to-image AI platform accessed primarily through Discord and its own web interface, and it remains one of the most discussed tools in AI image generation. Its underlying model changes on a regular schedule, and a review that doesn't say which version it's describing is not much use to someone deciding whether to sign up today.
As of September 2026, Midjourney's default model version is V8.2, which became the default on July 24, 2026. The previous version of this page predated the entire V8 series and never named a specific model version at all — a real gap for a review that ranks for "Midjourney review." This update rebuilds the page around the current release: what V8.2 actually changed, what happened to V6- and V7-era features along the way, and where the version lineup stands today.
Not an affiliate link.
This is the question the old version of this page never answered, and it's the one that matters most if you're comparing Midjourney to other generators or trying to figure out why a feature you read about isn't where you expect it. Midjourney has shipped several distinct main-line versions, each with its own default window:
If you subscribed a year ago and haven't checked since, you are almost certainly not on the version you think you're on. To set a version yourself, add --v # to the end of a prompt in the Imagine bar or in Discord, set a default version in the web settings panel, or use the settings command directly in Discord.
Alongside the speed increase, V8.1 introduced HD images — higher-resolution 2K output (2048px) generated directly, without a separate upscaling pass. You can toggle HD in the Version section of the web settings panel, or control it directly with the --hd and --sd parameters. On the web interface, a "Run batch as HD" button reruns a seed-locked prompt at HD resolution.
HD generation isn't free of trade-offs: it costs about 1.3 minutes of GPU time per job versus 0.8 minutes for a standard-definition job, per Midjourney's documentation. It's also worth knowing that using Pan, Zoom Out, or Edit/Vary Region on an HD image downscales the result back to standard definition — you'd need to run an Upscale option afterward to restore HD resolution. If you're budgeting compute time, that's a real cost to plan around rather than an edge case.
This is the part most "is Midjourney still good" searches are really asking about. Migrating from V6 or V7 to the current default isn't a pure upgrade — Midjourney's own feature compatibility chart shows a real list of features that simply aren't available on V8.1 or V8.2. Here's the chart, condensed:
| Feature | V6 | V7 | V8.1 / V8.2 |
|---|---|---|---|
| Character Reference (cref) | Yes | No | No (use Edit Model on V8.2) |
| Character Weight (cw) | Yes | No | No |
| Omni Reference | No | Yes | No (use Edit Model on V8.2; chart also notes V7) |
| Omni Reference Weight | No | Yes | No (uses V7) |
| Multi-Prompting | Yes | No | No |
| Quality Parameter | Yes (0.5, 1, 2) | Yes (1, 2, 4) | No |
| Turbo Mode | Yes | Yes | No |
| Draft Mode | No | Yes | No |
| Niji Version | --niji 6 | --niji 7 | No |
| Seed Parameter | Yes | Yes | Yes (~99% identical) |
| Max Aspect Ratio | 14:1 | 14:1 | 14:1 (4:1 for HD) |
| HD Images (2048px) | No | No | Yes |
The pattern is clear once it's laid out: several V7-only features — Omni Reference, Omni Reference Weight, and Draft Mode among them — did not carry forward into the V8 lineup, and Omni Reference is specifically flagged in Midjourney's own chart as still requiring V7 rather than working natively on V8. Several V6-only features, including Character Reference, Character Weight, Multi-Prompting, the Quality parameter, and Turbo Mode, are also absent from V8.1 and V8.2. If your workflow depended on the old cref or Omni Reference controls, those parameters do not carry forward as-is on default V8.2 — but Midjourney documents Edit Model as the replacement for Character Reference, Omni Reference, and Retexture. Treat it as a workflow migration, not a silent deletion with no path forward. For a dedicated side-by-side of Midjourney V8 vs V7: what changed, including which controls moved and which stayed, see that companion post.
Midjourney's official Version documentation (re-verified September 10, 2026) states that V8.2 features a new Edit Model, which replaces Omni Reference, Character Reference, and the Retexture tool. The compatibility chart marks several of those older reference/retexture paths as usable via Edit Model on V8.1/V8.2 rather than as the original cref/omni/retexture controls.
Practical takeaway: those features are not simply "gone with no path forward" on the current default. The legacy cref / Omni Reference / Retexture workflows do not run the same way on V8.2, but Midjourney documents Edit Model as the replacement path for those jobs. If your old prompts or Discord habits still call cref or Omni Reference directly, expect to relearn the Edit Model workflow rather than assuming the old parameters still apply on default V8.2.
It isn't all subtraction. Per Midjourney's documentation, the V8 lineup still supports Variations, Upscalers (for standard-definition images), Pan, Zoom Out, Remix, Personalization, Moodboards, the Editor, Style Reference, Style Reference Codes, Style Weight, Image Prompts, Image Weight, the No parameter, Chaos, Raw, Stylize, Tile, Weird, the experimental --exp flag, Conversational Mode, Relax Mode, and Fast Mode. Note that Pan, Zoom Out, and the Editor specifically run on V6.1 under the hood even when you're working in a V8 session — worth knowing if your results look like they're behaving slightly differently from the rest of your V8 generations.
Niji is a distinct series of Midjourney models, developed with Spellbrush and focused on Eastern and anime aesthetics and illustrative styles, with its own dedicated website and Discord community. The current version, Niji 7, launched January 9, 2026, with a coherency boost, clearer fine details in areas like eyes and reflections, closer prompt following, and a flatter, more line-focused default look.
The limitation worth flagging plainly: per Midjourney's own feature chart, Niji Version support is listed only for V6 (--niji 6) and V7 (--niji 7) — V8.1 and V8.2 show no Niji support at all. Since V8.2 is the current default, that means niji isn't available through the default Midjourney experience right now. We cover this model, its trade-offs, and workarounds in a full separate write-up: our niji-journey review.
Midjourney continues to fit well for designers, illustrators, marketers, and content creators who want a high degree of stylistic control and are comfortable iterating through prompts, Remix, and reference images rather than expecting one-shot precision. It's a weaker fit right now for anyone who still depends on the old Character Reference or Omni Reference workflows without adapting — those legacy controls don't run the same way on default V8.2, though Midjourney documents Edit Model as the replacement path — or for anyone specifically wanting the Niji anime aesthetic without leaving Midjourney's default V8 experience.
Midjourney's official documentation used for this review does not publish current subscription pricing — that information lives on Midjourney's own site and changes independently of model versions. Check Midjourney's official pricing page (not an affiliate link) for current plans before subscribing, rather than relying on a number quoted in any review, including earlier versions of this one.
Not an affiliate link.
Midjourney remains distinct from tools like DALL-E, Stable Diffusion, and Adobe Firefly mainly in its depth of prompt-and-parameter control — Style Reference, Chaos, Stylize, and Weird give you levers those tools don't expose in the same way — and in its community-driven Discord culture. Where Midjourney has genuinely changed ground in 2026 is version churn: three main-line default versions (V7, V8.1, V8.2) and a niche one (V8.0 alpha) rotated through in under eighteen months, each with a different feature set. That pace of change is itself worth factoring into any comparison — a generator that's simpler to use consistently may matter more than one with a longer feature list that shifts under you every few months. For a fuller side-by-side across tools, see our Best AI Image Generators roundup.
Midjourney is for still images; it does not do voiceover. If the next job in the same creative stack is AI voice, ElevenLabs is the tool this site already covers. That is a different job than image generation — not a Midjourney alternative.
Affiliate link.
V8.2, which became the default version on July 24, 2026. Per Midjourney's documentation, it's focused on aesthetics, image quality, and Personalization, and produces images the docs describe as "more creative, bold, sophisticated, and edgy."
The legacy Character Reference (cref) and Omni Reference controls do not run the same way on default V8.2. Per Midjourney's Version documentation (re-verified Sep 10, 2026), V8.2's Edit Model replaces Omni Reference, Character Reference, and the Retexture tool — so the jobs those features covered have a documented replacement path rather than disappearing with no alternative. The compatibility chart still marks the old cref/Omni controls as unavailable as-is on V8.1/V8.2 (Omni was a V7 feature; cref was V6). If you relied on those parameters directly, plan on learning Edit Model rather than pasting old cref/omni syntax into V8.2.
Yes. Midjourney's documentation describes V8.1, released April 14, 2026, as its fastest model so far, with standard jobs rendering roughly 4-5x faster than earlier versions, along with better prompt reading and detail retention.
HD images are 2048px outputs introduced with V8.1, generated directly without a separate upscale step — toggle them via the Version section of the web settings panel or the --hd/--sd parameters. They cost more compute: about 1.3 minutes of GPU time per job versus 0.8 minutes for standard definition. Using Pan, Zoom Out, or Edit/Vary Region on an HD image downscales it back to standard definition; an Upscale option restores HD resolution afterward.
No. Midjourney's feature compatibility chart lists Niji Version support only for V6 (--niji 6) and V7 (--niji 7); V8.1 and V8.2 show no Niji support at all. Since V8.2 is the current default, Niji isn't reachable through the default Midjourney experience — see our niji-journey review for how to use it separately.
For the full breakdown of exactly what changed between model generations, see our Midjourney V8 vs V7: What Changed post. If anime and illustrative styles are your focus, our niji-journey review covers that separate model lineage in depth. For the wider field of image generators, our Best AI Image Generators roundup compares Midjourney against the alternatives, and our AI Tools Starter Guide is a good starting point if you're new to evaluating these tools at all.