MidJourney launches V7 amid fierce competition in AI image generation market

The AI image generation firm MidJourney has launched its latest flagship model V7, which aims to accelerate creative workflows with a new ultra-fast ‘Draft Mode’ for rapid ideation at lower cost. The new model also introduces default personalization designed to tailor outputs to individual user tastes.

These user-centric upgrades arrive as the AI image generation firm navigates a fiercely competitive landscape increasingly defined by viral adoption cycles. Although the technology has been in the mainstream for years, AI image generators are arguably one of the hottest new technologies of 2025. The field has shifted dramatically since the 2023 launch of OpenAI’s DALL-E 3. In March, OpenAI’s latest image generation feature demonstrated the explosive viral potential of the technology when it triggered the so-called “Ghibli effect.” The company’s CEO Sam Altman noted that the company “added one million users in the last hour,” comparable to adding one million users in five days after ChatGPT’s initial launch, as Reuters reported. Users flooded social media with Studio Ghibli-style (maker of films ranging from My Neighbor Totoro to Spirited Away) transformations of photos and memes. The trend became so popular that OpenAI temporarily curbed use as Altman noted that its “GPUs were melting” from the demand surge.

From an R&D perspective, v7 offers concrete technical upgrades over v6 with rendering speeds roughly 20–30% faster. Its architecture combines multimodal and parallel diffusion transformer blocks scaled to 12 billion parameters. Consequently, that supports more efficient processing while maintaining high-quality output. The model also boasts improved multilingual capabilities that process prompts in various languages more effectively than previous versions. MidJourney founder David Holtz called the model its “most intelligent, aesthetically pleasing, and coherent model to date.” It offers improved coherence for complex elements such as human bodies, hands, and detailed objects, though user testing was mixed in these areas.

Based on the additional announcements information, here’s an HTML table summarizing the key aspects of MidJourney V7 that would enhance the article:

Category	Details
Core Improvements	Smarter text prompt interpretation Enhanced image prompt handling Higher image quality with improved textures Better coherence for bodies, hands, and objects Rendering speeds 20–30% faster than V6
Draft Mode	10x faster rendering speed 50% lower cost than standard jobs Conversational interface for quick iterations Voice input capability Compatible with “–repeat” and permutations Lower quality but consistent aesthetics Can be activated with “–draft” parameter
Personalization	First MidJourney model with default personalization Requires about 200 image ratings to establish profile Can be toggled on/off at any time Improves prompt interpretation based on user preferences
Available Modes	Turbo: Prioritizes speed (2x cost of standard V6 job) Relax: Slower but same quality (standard cost) Draft: Ultra-fast iteration (half the cost) Standard: Still being optimized (coming soon)
Current Limitations	Upscaling falls back to V6 models Editing features fall back to V6 models Retexturing falls back to V6 May require different prompting styles than V6
Upcoming Features	New features every 1–2 weeks for next 60 days New V7 character and object reference system Standard speed mode optimization Dedicated V7 upscaling and editing tools

AI image generators have emerged as one of the hottest new technologies of 2025, attracting millions of new users this year alone. The field has shifted dramatically since the 2023 launch of OpenAI’s DALL-E 3, with Black Forest Labs’ Flux models and Google’s Imagen3 displacing DALL-E 3, which saw its usage share plummet by nearly 80% between early 2024 and 2025. Other organizations in the market range from Stable Diffusion to Meta.

Before the launch, on April 2, the Midjourney team announced the cessation of its months-long “RELAX-ATHON,” which offered unlimited Relax mode use, to support adjustments to its server clusters in preparation for the v7 model launch.