Go to year: 2022 • 2023 • 2024 • 2025
Introduction
Now most models are good at text and aren't as prone to mangling human anatomy. A major addition this year included multiple models for editing images with text prompts.
Prompts and project details can be found at the bottom of the article. High resolution versions of the comparison image grids are in this article's attachment.
The Models
Lumina 2.0
January 2025
Lumina is a more lightweight release than most that will follow it this year. It seems like a marginal improvement over base SDXL in some ways, but can't compete with most current models. It is better at SDXL at following instructions, but it still struggles with anatomy and text.
Lumina 2
✅free download (GitHub | Hugging Face: all-in-one, diffusion variant)
Firefly Image 4
April 2025
Adobe released Firefly Image 4 and Firefly Image 4 Ultra. Based on my tests, the Ultra version seems to heavily favor realism over stylized images. Overall, it seems to be a big improvement over Firefly 3, but probably not worth choosing unless you're already in the Adobe ecosystem.
Firefly Image 4
❌free download
Firefly Image 4 Ultra
❌free download
GPT1 Image
April 2025
GPT1 has great prompt adherence, is good at text, and can make aesthetically pleasing images even when only given minimal prompts.
GPT1 Image
❌free download
HiDream-I1
April 2025
HiDream had great prompt adherence, is good with text, and leans toward stylized images. Both the Dev and Full versions are available to run locally.
HiDream-I1 Dev
✅free download (CivitAI | Hugging Face: fp8/bf16, gguf)
HiDream-I1 Full
✅free download (CivitAI | Hugging Face: fp8/bf16)
Midjourney 7
April 2024
Midjourney's new model emphasizes personalization that can adapt to the user's preferences. This means you might get images that look very different from mine with the same prompts.
Midjourney 7
❌free download
Seedream 3
April 2025
Seedream 3 is ByteDance's text-to-image model. It's good with text and aesthetics, but sometimes seemed to prioritize style over following my instructions. This version is pretty good, but it's going to have a couple more updates in only a few months.
Seedream 3
❌free download
Google imagen 4 (reg and ultra)
August 2025
This model from google seems to favor stylized images. It's good at prompt adherence and great with text.
Google Imagen 4
❌free download
Google Imagen 4 Ultra
❌free download
Nano Banana
August 2025
A.K.A Gemini 2.5 Flash Image/Gemini 3 Pro Image. Another model under the Google umbrella. It's good at prompt adherance, text, and aesthetics. It can be used to generate images directly and to edit existing images.
Nano Banana
❌free download
Qwen Image
August 2025
Qwen is currently a community favorite with an Apache license. It's great at prompt adherence, text, anatomy, and styles. It's resource intensive, but variations have been released that can be run on lower vram graphic cards.
✅free download (CivitAI | Hugging Face: fp8/bf16, gguf, nunchaku)
Seedream 4.0
September 2025
ByteDance's second release of Seedream this year, but not its last.
Seedream 4.0
❌free download
Hunyuan Image 2.1
September 2025
Hunyuan can be run locally but it is resource-intensive. It is good with text and prompt adherence. It can create aesthetically pleasing images, but the images tended to be more basic without extra guidance. Without a refiner, this model tends to be blurry and lack fine details.
✅free download (Hugging Face: official, gguf)
Kandinsky 5
November 2025
Kandinsky is a Russian model, and if you use it long enough that may become obvious since it sometimes randomly adds Russian text to images. It's one of the more inconsistent models I tested; sometimes it's great at prompt adherence and other times it ignores basic requests and does its own thing.
✅free download (Hugging Face)
Z Image Turbo
November 2025
Z-Image Turbo is one of the few new models released this year that can be run on lower vram and still offer significant improvements over the older Stable Diffusion models. It's good with text, realism, and is fairly good at prompt adherence (my experience was good but not as good as some of the other current models).
✅free download (CivitAI | Hugging Face: bf16, fp8, gguf)
Flux 2
November 2025
The latest in the Flux lineup has been released, and it's quite resource hungry. I haven't had much time to experiment with it yet.
Flux Dev 2.0
✅free download (CivitAI | Hugging Face: fp8, gguf)
Flux Pro 2.0
❌free download
Seedream 4.5
November 2025
And yet another Seedream release. I haven't used it much, but it looks like it has a focus on creating stylized images.
Seedream 4.5
❌free download
Project Details
Disclaimer
I'm not an insider with special access to anything or a programmer who understands how all this works under the hood. I took some time to research, but this is from information found online and I can't guarantee everything is accurate. This is a work in progress; I'm still working on filling in missing information.
Also note that this is only a comparison of base models. Some models can produce significantly better images by using trained checkpoints, styles, presets, or detail enhancers.
Criteria
Must still be publicly accessible in 2025 without a complicated setup.
I'm trying to keep no more than two of each release: a "top of the line" version and the smaller version released for consumers. I've left out variants like those for turbo, editing, and less resource intensive versions.
Process
I chose 15 prompts that show a variety of photo realism, art styles, people, animals, objects, specific instructions, open-ended short prompts, text, and abstract concepts.
All images come from the first generation set and I never picked from more than 1-4 images.
When possible, I used images from the same seed which can show differences between minor versions of the same model.
I used the recommended settings for each model or the default offered online.
I didn't use additional styles or presets.
Prompts
african hydropunk princess
artificial intelligence
astronaut exploring an alien planet
overhead view of a breakfast plate with eggs, toast, strawberries, coffee, and a fork
exterior of a cafe watercolor painting
person wearing cyberpunk accessories in a high tech neon city
druid man character design
ethereal fairy in the style of oil painting
graphic design logo with fennec fox and succulents and text "Desert Design"
man and a woman in love
photo of a deer in an enchanted forest with cinematic lighting
Photo portrait of a woman with long black curly hair in natural light. She's wearing a fashionable purple blouse, a gold necklace with a locket, and hoop earrings. Bokeh background.
pixel art city street scene with shops and pedestrians at night
red potion bottle with text "health" on the left, blue potion bottle with text "mana" in the middle, green potion bottle with text "poison" on the right, on a wooden table in a dark alchemist's laboratory, in the style of a detailed digital painting
woman lying on the grass