Flux 1.1 Pro Ultra can generate up to four times higher image resolutions than the standard mode.
Black Forest Labs (BFL) has released a brand new Ultra mode for the Flux 1.1 Pro AI image model. This new mode is capable of generating up to four times higher image resolutions than the standard mode. The boosted resolution does not compromise speed, as it can generate super high-quality photos in under 10 seconds.
Take a look at the cover image for this article. That stunning photo was generated using the new Flux 1.1 Pro in Ultra mode, with the following prompt:
A breathtaking, majestic snow-capped mountain peak is enveloped in the warm, golden glow of a setting sun, casting enchanting shadows and illuminating the rugged terrain with a serene beauty.
I can’t help but be amazed by how much detail this new mode can produce in a single frame.
According to BFL, unlike many high-resolution models that experience significant slowdowns at higher resolutions, their benchmarks show sustained fast generation times — over 2.5 times faster than comparable high-resolution offerings.
This level of performance is crucial for creators who need to produce high-quality images on demand without waiting minutes for each one to render.
In case you’re wondering how they arrived at this benchmark for the image above, BFL used an ELO score to rank Flux against its competitors. The text-to-image leaderboard ranks these models by their ELO scores, determined by pairwise image comparisons labeled by visitors to the Image Arena. This system is effective because it allows real users to evaluate the output quality of each model, leading to more accurate comparisons based on practical, real-world conditions.
The ELO rating system is commonly used to rank chess players but has been adapted here to evaluate and compare image generation models.
In terms of image quality, Flux surpasses the likes of Midjourney, Dall-E 3, and Stable Diffusion.
If you want to know how they evaluate these image models, check out Artificial Analysis’ methodology page.
Check out some example images generated with Flux 1.1 Pro Ultra mode in Flux Labs AI.
Under the Flux model dropdown menu, make sure to set it to “Ultra High-quality” before generating your image.
Prompt: A hidden valley, with wildflowers carpeting the ground and cascading waterfalls
Prompt: a close-up of a blueberry surrounded by a splash of dark blue water and tiny leaves and bubbles, across a blurred background
Prompt: the serene interior of a cave, where a pool of water is nestled amidst an abundance of greenery. Sunlight filters through the cave opening, casting a soft glow on the surrounding foliage and the tranquil water below
Prompt: a majestic snow-capped mountain peak bathed in a warm glow of the setting sun
Prompt: Image centered on three handsome, trendsetting Chinese male singers, their faces beams with radiant smiles as they joyfully belt out tunes in a KTV room filled with bright, neon-infused light. Render the scene through a fisheye lens from a bird’s-eye view, distorting the space and adding to the lively ambiance. Immerse the room in a verdant, green hue that complements the neon accents, creating a futuristic and captivating backdrop. The lighting should be vivid and radiant, illuminating the singers and accentuating their good looks and the sheer delight of the moment.
On Replicate, this model is available at a competitive price of $0.06 per image or 16 images per $1.
Here’s the pricing comparison with other Flux models:
You can also use BFL’s API (currently in beta) or Freepik if you want to integrate it into your app. For me, I use Replicate and Fal for my AI SaaS applications.
Aside from the Ultra mode, BFL also released a Raw mode for Flux 1.1 Pro. This new mode is for creators seeking authenticity. It captures the genuine feel of candid photography—less synthetic, more natural aesthetic.
Compared to other text-to-image models, raw mode significantly increases diversity in human subjects and enhances the realism of nature photography
Here are some examples:
I don’t know about you, but these images are so realistic it could easily fool someone to believe they’re real.
Overall, I’m genuinely impressed with how fast BFL is moving forward and releasing such powerful image models.
In case you didn’t know, the engineers behind BFL are the same team that created the diffusion model technology, which became the foundation of Stability AI’s popular Stable Diffusion models. That background is clear when you see what they’ve achieved with Flux 1.1 Pro Ultra mode.
It only took me a few hours to integrate Flux 1.1 Pro Ultra mode into Flux Labs AI, and so far, I’m super satisfied with the results. I’ve been using it so much that I decided to cancel my Midjourney subscription; Flux models are now my go-to for image generation in my projects. The quality, speed, and detail they offer are really impressive.
What I’m most excited about now is the possibility of custom Loras based on the new Flux models, which could open up even more creative possibilities.
‍
Software engineer, writer, solopreneur