Discover legacy tools and features that shaped the Midjourney evolution.
If you’re curious about the older versions of Midjourney and the features they offer, you’ve come to the right place. As we keep moving forward with the latest updates, we also want to acknowledge and explore the tools and options from previous models that helped shape Midjourney to what it is today.
Legacy Versions
Niji 6
The Niji 6 model, released on June 7, 2024, features improved rendering of Japanese text, handling short Japanese kana text effectively and even rendering simple Chinese characters. It offers better image detail, particularly in elements like the structure of anime eyes, and addresses certain minor image artifacts that affected a small number of previous generations.
Version 5.2
Version 5.2 was released in June 2023. It was the default model from June 22, 2023 to February 14, 2024. Version 5.2 produces more detailed, sharper results with better colors, contrast, and compositions. It also has a slightly better understanding of prompts than earlier models and is more responsive to the full range of the --stylize parameter.
Version 5.1
Version 5.1 was released on May 4th, 2023. It was the default model from May 3, 2023 to June 22, 2023. Version 5.1 has a stronger default aesthetic than earlier versions, making it easier to use with simple text prompts. It also has high coherency, excels at accurately interpreting natural language prompts, produces fewer unwanted artifacts and borders, has increased image sharpness, and supports advanced features like repeating patterns with --tile.
Version 5
Version 5 was released in March 2023. It was the default model from March 30, 2023 to May 3, 2023. Version 5 produces more photographic generations than the V5.1 model, and produces images that closely match the prompt but may require longer prompts to achieve your desired aesthetic.
Niji 5
The Niji 5 model was released in April 2023. It can be fine-tuned with specific --style parameters including:
• --style cute creates charming and adorable characters, props, and settings.
• --style expressive has a more sophisticated illustrated feeling.
• --style original uses the original Niji Model version 5, which was the default before May 26th, 2023.
• --style scenic makes beautiful backgrounds and cinematic character moments in the context of their fantastical surroundings.
Default Parameter Values
| Aspect Ratio | Chaos | Quality | Seed | Stop | Stylize | |
|---|---|---|---|---|---|---|
| Default Value | 1:1 | 0 | 1 | random | 100 | 100 |
| Range | any | 0-100 | 0.25, 0.5, 1 | 0-4294967295 | 10-100 | 0-1000 |
Upscalers: Version 5 features the Upscale (2x) and Upscale (4x) tools, which are not compatible with Pan or tile.
-
Version 4
Version 4 was released in November 2022. It was the default model from December 20, 2022 to March 30, 2023. Version 4 is an entirely new codebase and brand-new AI architecture (compared to previous versions) designed by Midjourney and trained on the new Midjourney AI supercluster. Version 4 has more knowledge of creatures, places, objects, and more than previous versions. It's much better at getting small details right and can handle complex prompts with multiple characters or objects. Version 4 supports advanced functionality like image prompting and multi-prompts, and has very high coherency and excels with Image Prompts.
Version 4 has three slightly different "flavors" with slight tweaks to the stylistic tuning of the model. Experiment with these versions by adding--style 4a,--style 4b, or--style 4cto the end of a V4 prompt. 4c is the default and support aspect ratios up to 1:2 or 2:1. 4a and 4b only support 1:1, 2:3, and 3:2 aspect ratios.Niji 4
The first Niji model, Niji 4, was released on December 20, 2022.
Default Parameter ValuesAspect Ratio Chaos Quality Seed Stop Style Stylize Default Value 1:1 0 1 random 100 4c 100 Range 1:2–2:1 0-100 0.25, 0.5, 1 0-4294967295 10-100 4a, 4b, 4c 0-1000
Upscalers: Earlier Midjourney model versions start by generating a grid of low-resolution image options for each Job. You can use a Midjourney upscaler on any of these images to increase the size and add additional details. The table below shows image sizes for 1:1 ratio images. For more information view our Upscalers article.Model Version Starting Grid Default Upscale Detail Upscale Light Upscale Beta Upscale Version 4 512 x 512px 1024 x 1024px 1536 x 1536px 1536 x 1536px 2048 x 2048px Niji 4 512 x 512px 1024 x 1024px 1024 x 1024px 1024 x 1024px 2048 x 2048px
-
The
--testand--testpmodels were introduced in August 2022 as part of an effort to gather community feedback and improve upon new features. These test models allow members to explore upcoming capabilities, and they can be used alongside the--creativeparameter for more varied compositions. When using these models, keep in mind that they support--stylizevalues ranging from 1250 to 5000, offering strong artistic influence.There are some limitations with the test models. They do not support multi-prompts or image-prompts and have a maximum aspect ratio of 3:2 or 2:3. When the aspect ratio is 1:1, they generate two initial grid images, but with other aspect ratios, they produce only one initial grid image. Additionally, in crafting prompts, note that words at the beginning of your prompt may hold more weight than those at the end. These unique characteristics make the test models an exciting option for experimental image creation.
Upscalers: Initial grid images are 512 x 512px and the only option to upscale is Beta Upscale, which creates a 2048 x 2048px image when using 1:1 aspect ratio. For more information view our Upscalers article.
-
Version 1
Version 1 was the default model from February 2022 to April 2022. Version 1 is very abstract and painterly with low coherency.
Version 2
Version 2 was the default model from April 2022 to July 2022. Version 2 is creative, colorful, and painterly with low coherency.
Version 3
Version 3 was the default model from July 2022 to November 2022. Version 3 generates highly creative compositions with moderate coherency.
V3 Legacy Parameter:--sameseedSeed values create a single large random noise field applied across all images in the initial grid. When--sameseedis specified, all images in the initial grid use the same starting noise and will produce very similar generated images.High Definition
--hdis an early alternative model that generates busy, detailed, and abstract images with low coherency.
Upscalers: Earlier Midjourney model versions start by generating a grid of low-resolution image options for each Job. You can use a Midjourney upscaler on any of these images to increase the size and add additional details. The table below shows image sizes for 1:1 ratio images. For more information view our Upscalers article.Model Version Starting Grid Default (Detail) Upscale Light Upscale Beta Upscale Max Upscale Version 1-3 256 x 256px 1024 x 1024px 1024 x 1024px 1024 x 1024px 1664 x 1664px hd 512 x 512px 1536 x 1536px 1536 x 1536px 2048 x 2048px --
Legacy Parameters
|
High Definition Model --hd |
Test Models --test --testp |
Creative --creative |
|
Sameseed --sameseed |
Uplight --uplight |
Upbeta --upbeta |
|
Stop --stop |
Style Tuner Codes --style code |
Deprecated Parameters
--width and --w (replaced with Aspect Ratio)
--height and --h (replaced with Aspect Ratio)
--fast (replaced with Quality)
--vibe (now known as model version 1)
--upanime
--hq
--newclip
--nostretch
--old
Parameter Version Compatibility
For information on the current default version (V7), please visit our Version article.
| Affects Initial Images | Affects Variations & Remix | V6 | V5 | V4 | V3 | Test & Testp | Niji 4 | Niji 5 | |
|---|---|---|---|---|---|---|---|---|---|
| Max. Aspect Ratio | any | any | 1:2 / 2:1 | 5:2 / 2:5 | 3:2 / 2:3 | 1:2 / 2:1 | any | ||
| Chaos | |||||||||
| Image Weight | 0-3 default 1 |
0.5-2 default 1 |
0.5-2 default 1 |
any default 0.25 |
0.2-2 default 1 |
||||
| No | |||||||||
| Quality | 0.5-2 | 0.25-1 | 0.25-1 | 0.25-2, 5 | 0.25-1 | 0.25-1 | |||
| Repeat | |||||||||
| Seed | |||||||||
| Sameseed | |||||||||
| Stop | |||||||||
| Style | raw | raw | 4a, 4b | expressive, cute, scenic, original | |||||
| Stylize | 0-1000 default 100 |
0-1000 default 100 |
0-1000 default 100 |
625-60000 default 2500 |
1250-5000 default 2500 |
0-1000 default 100 |
|||
| Tile | |||||||||
| Weird | |||||||||
| # of Initial Images | -- | -- | 4 | 4 | 4 | 4 | 2 or 1 if custom aspect ratio | 4 | 4 |
Legacy Upscalers
In previous Midjourney model versions, images generated at smaller sizes. You can still enlarge these images using a legacy upscaler. These tools not only increase the size but also add more details. There are various legacy upscalers you can choose from for images made with earlier models. Keep in mind, using a legacy upscaler will use GPU minutes.
Version 5
Midjourney model versions 5.x produce 1024 x 1024 pixel images. You can then use the Upscale (2x) or Upscale (4x) tools to increase the size of your image to 2048 x 2048 pixels or 4096 x 4096 pixels respectively. V5 upscale tools are not compatible with the Pan tool or the tile parameter.
Versions 4 & Earlier
Each of these upscalers changes your image in different ways. All sizes are in pixels and for square aspect ratios.
| Version | Initial Image | V4 Default Upscale | Light Upscale | Beta Upscale | Detail Upscale | Max Upscale** |
|---|---|---|---|---|---|---|
| V4 | 512 x 512 | 1024 x 1024 | 1536 x 1536 | 2048 x 2048 | 1536 x 1536 | -- |
| V1-V3 | 256 x 256 | -- | 1024 x 1024 | 1024 x 1024 | 1024 x 1024* | 1664 x 1664 |
| Niji | 512 x 512 | 1024 x 1024 | 1024 x 1024 | 2048 x 2048 | 1024 x 1024 | -- |
| test / testp | 512 x 512 | -- | -- | 2048 x 2048 | -- | -- |
| hd | 512 x 512 | -- | 1536 x 1536 | 2048 x 2048 | 1536 x 1536* | -- |
* = the default upscaler for each Midjourney version model.
** Max Upscale is an older resource-intensive upscaler and is only available when in Fast Mode.
When using legacy versions, the --uplight or --upbeta parameters in your prompt will turn the U# buttons in Discord into Light Upscale or Beta Upscale shortcut buttons.
Style Tuner
The Style Tuner was a Discord feature in Midjourney version 5.2 and was one of the first steps toward developing Personalization, Moodboards, and the Style Explorer. The /tune command no longer works, and new Style Tuners cannot be created.
You can still access previously created Style Tuners using the /list_tuners command, and existing --style code parameters can still be used.
Remaster
Remaster is a Discord button for upscaled images generated using legacy versions of Midjourney. It automatically remixes your prompt into version 5.2. Remaster any previously upscaled job by clicking the 🆕 Remaster button beneath the upscaled image.
Stop
Stop is a parameter compatible with versions 6 and earlier. Using --stop allows you to end the image creation process before it's fully completed, resulting in softer, less detailed images.
The default stop value is 100, meaning the image completes fully unless you specify otherwise. You can choose a value between 10 to 100. This value represents the percentage of the process you'd like to stop at.
Rooms
Rooms were shared spaces on the Midjourney website where members could chat and generate creations together. They were discontinued on February 26, 2026. For events, collaboration, and socializing check out the Midjourney Discord server.