--- title: "Midjourney says V9 could double its cost per image" description: "The self-funded lab is asking subscribers to accept twice the per-image cost for quality, the same week MiniMax open-sourced a 4K model whose license skips the US and the EU. The answer will price every credit-metered image AI, Adobe's included." publisher: "The Inference" section: "Model" published: 2026-11-12T18:24:30.954Z modified: 2026-11-12T18:24:30.954Z canonical: https://theinference.org/article/midjourney-says-v9-could-double-its-cost-per-image language: en keywords: "Midjourney, MiniMax, Image Generation, Model Pricing, Open Weights" --- # Midjourney says V9 could double its cost per image > The self-funded lab is asking subscribers to accept twice the per-image cost for quality, the same week MiniMax open-sourced a 4K model whose license skips the US and the EU. The answer will price every credit-metered image AI, Adobe's included. ## Subscribers get the bill for a bigger V9 At [Midjourney](https://theinference.org/markets/companies/midjourney)'s office hours on November 11, founder [David Holz](https://theinference.org/markets/people/david-holz) scoped V9 for release within 30 days, bundling a new base model, a new upscaler and a rebuilt style-reference system [1]. The cost per image could double, no final price is set, and a stronger version of the new Thinking mode could cost about ten times more [1]. "I think I want to favor clearly superior quality," Holz said, in remarks translated from the [session notes](https://geniart.fr/midjourney/office-hours/midjourney-v9-texte-temps-reel/) [1]. Real-time generation, which would let users iterate cheaply before a final render, may not make V9 at all [1]. The bill lands on subscribers, who pay $10 to $120 a month plus $4 an hour for extra fast GPU time [2], and on Midjourney itself: the lab calls itself community-funded, self-funded and 60 people strong [3], profitable since 2022 by its founder's account [4]. A mispriced V9 lands on its own books, with no investor to absorb the miss [2][3][4]. ## The free 4K rival is fenced out of Midjourney's markets The same week, MiniMax, a Shanghai company listed in Hong Kong since January [5], released the weights of H3-Image, whose 33.1-billion-parameter image transformer renders about 16 megapixels directly, with no upscaler, in six sampling steps and up to 16 reference images [6]. The launch post claims a 4K image in under 10 seconds [7]. [The license](https://huggingface.co/MiniMaxAI/MiniMax-H3-Image/raw/main/LICENSE) does the fencing: it applies everywhere except the European Union, the United Kingdom, South Korea and the United States, demands MiniMax's prior written permission above $1 million in annualized revenue or for any hosted service, and bars outputs from leaving the permitted territory [8]. MiniMax drew that million-dollar fence where Holz drew Midjourney's years ago [13]. The free download arms developers in Japan and other permitted countries [11]; in Midjourney's home markets the rival is [MiniMax's paid API](https://theinference.org/article/anthropic-s-16-3-billion-half-year-meets-minimax-s-0-15-rate-card), [about $0.022 a 1K image and $0.069 at 4K](https://cryptobriefing.com/minimax-h3-image-open-weight-4k-model/) [9], beside [OpenAI](https://theinference.org/markets/companies/openai)'s Images 2.6, shipped November 4 with 50 percent faster generation [10]. *Chart: **A 4K image now lists between 7 and 61 cents on rival APIs** Bar chart of listed API price for one 4K image: MiniMax H3-Image at $0.069, Google Nano Banana 2.1 at $0.113 and BFL FLUX 3 Image at $0.607.* *List API price for one 4K (4096x4096) image as reported in November 2026; the MiniMax figure is Standard mode. OpenAI bills Images 2.6 per token rather than per image, so it is excluded. Sources: Crypto Briefing; MarkTechPost. [9][10]* ## Four billion images a week, and a seven-cent floor Quality still sells at volume: OpenAI says users now generate more than 4 billion images a week, up from 3 billion eight weeks earlier [10]. The floor still disciplines the ceiling: MiniMax's own chart puts [median 4K text-to-image at 10.3 seconds](https://xenospectrum.com/en/minimax-h3-image-4k-speed/) even in its fast tier, above the advertised claim, and no independent timing had been published as of November 11 [11][5]. ## Watch the 30-day clock on quality pricing V9 is due by about December 11, with Niji, the anime model, hoped within a week; Holz called both the lab's last big model releases targeted this year, with V10 and video framed for next year [1]. [Adobe](https://theinference.org/markets/companies/adobe) meters Firefly generation through [monthly credit allocations in paid plans](https://www.adobe.com/products/firefly.html) [12] and faces the same question at scale. If beauty still bills, every credit meter holds its price; if not, the floor is seven cents a 4K image [12]. ## Takeaway Midjourney is betting subscribers will pay up to twice as much per image for a better V9, at the same moment a free 4K rival is fenced out of its home markets and seven-cent 4K pricing sets a floor under every credit meter. ## Sources 1. [GeniArt, "Midjourney V9 : texte amélioré, temps réel encore incertain," notes from Midjourney Office Hours of 11 November 2026, published 12 November 2026](https://geniart.fr/midjourney/office-hours/midjourney-v9-texte-temps-reel/) 2. [Midjourney documentation, "Comparing Midjourney Plans," accessed 12 November 2026](https://docs.midjourney.com/hc/en-us/articles/27870484040333-Comparing-Midjourney-Plans) 3. [Midjourney, homepage ("community-funded research lab of 60 people"; "lean, self-funded, distributed team"), accessed 12 November 2026](https://www.midjourney.com/) 4. [Wikipedia, "Midjourney," citing Thomas Claburn, The Register, 1 August 2022 (Holz on profitability), accessed 12 November 2026](https://en.wikipedia.org/wiki/Midjourney) 5. [Tech Times, "MiniMax Releases Open-Weight Image Model H3-Image: Only Pricier Highspeed Tier Nears 10-Second 4K," 12 November 2026](https://www.techtimes.com/articles/330529/20261112/minimax-releases-open-weight-image-model-h3-image-only-pricier-highspeed-tier-nears-10-second-4k.htm) 6. [MiniMaxAI, MiniMax-H3-Image model card, Hugging Face, accessed 12 November 2026](https://huggingface.co/MiniMaxAI/MiniMax-H3-Image) 7. [MiniMax, "MiniMax H3-Image: Bringing 4K Image Generation into the 10-Second Era," 11 November 2026](https://www.minimax.io/blog/minimax-h3-image) 8. [MiniMax-H3-Image Community License Agreement, license date 10 November 2026, Hugging Face](https://huggingface.co/MiniMaxAI/MiniMax-H3-Image/raw/main/LICENSE) 9. [Crypto Briefing, "MiniMax releases open-weight H3-Image model with 4K output on Hugging Face," 10 November 2026](https://cryptobriefing.com/minimax-h3-image-open-weight-4k-model/) 10. [MarkTechPost, "OpenAI Launches GPT-Image-2.6 Flare and Sunburst: Adds GIFs, SVG Export and 6-Second API Generation," 4 November 2026](https://www.marktechpost.com/2026/11/04/openai-gpt-image-2-6-flare-sunburst/) 11. [Xenospectrum (translated from the Japanese original), "MiniMax's H3-Image Generates 16-Megapixel Images Directly, but Its Speed Claims Need Context," checked 11 November 2026](https://xenospectrum.com/en/minimax-h3-image-4k-speed/) 12. [Adobe, "Adobe Firefly" product page, generative-credits FAQ, accessed 12 November 2026](https://www.adobe.com/products/firefly.html) 13. [Forbes, Rob Salkowitz, "Midjourney Founder David Holz On The Impact Of AI On Art, Imagination And The Creative Economy," 16 September 2022](https://www.forbes.com/sites/robsalkowitz/2022/09/16/midjourney-founder-david-holz-on-the-impact-of-ai-on-art-imagination-and-the-creative-economy/) --- Midjourney says V9 could double its cost per image — The Inference. Canonical: https://theinference.org/article/midjourney-says-v9-could-double-its-cost-per-image