MidJourney

Overview

Midjourney V7 represents a fundamental leap in AI image generation, introducing Draft Mode for 10x faster iteration at half the cost, a Photoshop-style layer editor, and Omni Reference for consistent character and object placement across scenes. Released April 3, 2025 with alpha testing announced April 4, V7 became the default model on June 17, 2025. Fast mode shipped May 2, 2025, followed by a performance optimization on May 29 that brought V7 approximately 40% faster than V6.1 (22 seconds in Fast mode vs 36 seconds, 9 seconds in Turbo vs 13 seconds). V7 delivers significantly higher image quality with richer textures and better coherence in complex details like bodies, hands, and objects, and features model personalization enabled by default, learning from your aesthetic preferences to better interpret your creative vision.

What's New

Draft Mode: 10x Faster Iteration

Draft Mode renders images at 10 times the speed of standard generation while costing half as much GPU time, making it ideal for rapid prototyping and creative iteration. When enabled on the web interface, the prompt bar transforms into a conversational mode where you can request changes like "swap out the cat with an owl" or "make it nighttime" and the AI will automatically manipulate the prompt and start a new job. Draft mode also supports voice input, allowing you to think out loud while images flow beneath you. While Draft images are lower quality than standard mode, the behavior and aesthetics remain consistent, making it a faithful way to iterate. When you find a composition you like, simply click "enhance" or "vary" to re-render at full quality. You can also use --draft explicitly in prompts for permutations or --repeat commands.

Advanced Image Editor with Layer Editing

V7 introduces a Photoshop-style multi-layer editor accessible to all membership tiers, unifying the experience for editing both Midjourney-generated and externally uploaded images. The editor supports layer-based composition, allowing you to import multiple images on separate layers to create complex collages and retextures. Key tools include smart selection for precise object removal with a single click, masking and cutting capabilities, and full layout control for arranging elements. The retexture feature lets you change textures and styles of individual image elements while maintaining structural coherence. This represents Midjourney's most significant step toward becoming a complete design tool rather than just a generation platform.

Omni Reference: Unified Character and Object Consistency

Omni Reference (--oref) replaces the previous Character Reference system with a more flexible approach that references structure and content aspects of source images simultaneously. Unlike the previous Character Reference which focused solely on facial features, Omni Reference maintains consistency across entire characters, objects, and compositional elements across multiple scenes. For style consistency, use Style Reference (--sref) in combination with Omni Reference. Note that Omni Reference costs 2x GPU time and is not compatible with certain editing operations like inpainting and outpainting. This feature is particularly powerful for storytelling, branding, and maintaining visual coherence in multi-image projects when working within the editor or generating new variations.

Model Personalization by Default

V7 is the first Midjourney model to have personalization enabled by default. Based on internal data, 85% of users prefer personalized results. Personalization requires unlocking through a 5-minute ranking process where you rate generated images to teach the AI your aesthetic preferences. Once enabled, the model adjusts its interpretation of your prompts based on what it has learned you find beautiful. You can toggle personalization on or off at any time, and the system continues learning from your interactions. This works in conjunction with Moodboards, which allow you to create visual reference boards that further guide the AI's output style.

Enhanced Image Quality and Coherence

V7 delivers noticeably higher image quality with richer textures and dramatically improved coherence in complex details. Bodies, hands, and objects of all kinds show significantly better anatomical accuracy and logical spatial relationships. The model excels at interpreting both text and image prompts with greater precision, producing results that more closely match creative intent. Textures appear more realistic and detailed, with improved rendering of materials like fabric, metal, glass, and organic surfaces. While Midjourney hasn't detailed the technical training approach, V7 empirically demonstrates better understanding of complex creative briefs including abstract concepts and multi-subject scenes.

Additional Parameters and Controls

V7 supports several parameters you can add to prompts for different behaviors. The --raw parameter provides more literal prompt interpretation without personalization. An experimental --exp parameter is available for testing new capabilities. Quality settings (--q) now accept values of 1, 2, or 4, each with distinct rendering behaviors and GPU cost implications. The model supports any aspect ratio, strong and subtle variations, remix features, pan, zoom, and tile parameters, giving granular control over generation quality and creative exploration.

Pricing & Plans

Midjourney V7 is available across all existing subscription tiers with no price increase. All plans include access to Draft Mode, the Advanced Editor, and Omni Reference.

Basic Plan - $10/month (or $96/year, saving 20%)

• 3.3 hours of fast GPU time per month (~200 images)
• Draft Mode access (double your image output at half the cost)
• Multi-layer editor with smart selection
• Omni Reference and personalization

Standard Plan - $30/month (or $288/year)

• 15 hours of fast GPU time
• Unlimited relax mode generations
• All Basic features
• Priority support

Pro Plan - $60/month (or $576/year)

• 30 hours of fast GPU time
• Unlimited relax mode
• Stealth mode (private images)
• All Standard features

Mega Plan - $120/month (or $1,152/year)

• 60 hours of fast GPU time
• Unlimited relax mode
• Stealth mode
• All Pro features
• Highest priority in queues

Additional GPU time can be purchased at $4/hour across all plans. V7 Turbo mode costs 2x standard GPU time (same as V6 Turbo), while Draft Mode costs 0.5x standard GPU time. Fast mode shipped on May 2, 2025 and received further performance optimizations on May 29, 2025.

Pros & Cons

Pros:

• Draft Mode provides 10x generation speed at half the cost for rapid iteration and brainstorming
• 40% faster than V6.1 in Fast and Turbo modes (22s vs 36s, 9s vs 13s) for standard generations
• Photoshop-style layer editor with smart selection now available to all membership tiers, not just premium
• Omni Reference maintains character and object consistency across multiple scenes better than previous Character Reference
• Significantly improved coherence in bodies, hands, and objects addresses longstanding AI art challenges
• Model personalization learns your aesthetic preferences for more intuitive prompt interpretation
• Richer textures and higher overall image quality compared to V6.1
• No price increase despite major feature additions and performance improvements

Cons:

• Draft Mode images are noticeably lower quality and require re-rendering at full quality for final use
• Omni Reference costs 2x GPU time and is incompatible with inpainting and outpainting operations
• Some expand features (Pan, Zoom Out) still use V6.1 models while V7 versions are being developed
• Layer editor doesn't yet match Photoshop's full capabilities compared to other AI image editing tools for advanced design work
• Requires minimum 5-minute ranking session to unlock personalization before first use
• V7 may require different prompting styles than previous versions
• Turbo mode costs 2x standard GPU time, making high-speed generation expensive for heavy users

Best For

• Content creators and marketers who need to rapidly prototype visual concepts with AI image generators and iterate on ideas before final production
• Digital artists and illustrators who want to maintain consistent characters or objects across multiple scenes for storytelling or branding
• Graphic designers who need to composite AI-generated elements with traditional design work using layer-based editing
• Social media managers producing high volumes of visual content who can leverage Draft Mode's speed and cost efficiency
• Creative professionals who value aesthetic control and want AI that learns their personal style preferences through personalization
• Professional photographers and retouchers who need to enhance or manipulate existing images using AI-powered editing tools

FAQ