Hands-on comparison of MiniMax H3-Image, GPT Image 2.6, and Seedream 5.5
The author used the same set of complex prompts to test MiniMax H3-Image, GPT Image 2.6, and Seedream 5.5 on Hailuo AI, ChatGPT, and Doubao. H3-Image's high-speed mode generates images in about 10-15 seconds, faster than the roughly 1 minute of the other two, but under complex instructions it showed issues such as clipping and failure to blend the face with the tanghulu; GPT Image 2.6 had the most accurate comprehension and modification, while Seedream 5.5 missed changes in local editing and took about 7 minutes. The article also mentions that MiniMax opened the H3-Image weights on the evening of November 10, priced at 50%-70% of mainstream tools in the same tier.
Why read this
AI score 64
The author used the same set of complex prompts to compare the three new image models side by side, detailing specific differences in speed, comprehension, and editing stability.
Topics
Body · AI translation
On the evening of November 10, MiniMax released and opened the weights of the H3-Image model, playing the "fast and cheap" card. The company claims it can generate a 16-megapixel high-definition image in just 10 seconds, at only 50%-70% of the price of mainstream tools in the same tier.
Before it, OpenAI and ByteDance had just handed in their homework.
On November 5, OpenAI launched Images 2.6GPT Image 2.6, continuing to push on speed and editing capabilities, claiming a 50% improvement in image generation speed over the previous generation, and adding features such as GIF animation generation and reusable reference images. On the API side, it introduced two models, GPT Image 2.6 Flare and Sunburst, with the latter focused on consistency in multi-turn editing.
ByteDance, meanwhile, released Seedream 5.5 on November 9, focusing on the stability of continuous editing and supporting precise local modifications.
All three have rolled out their own new models—just how big are the actual differences in performance? Beyond competing on generation quality, what else are they fighting over?
01 OpenAI leads, MiniMax wins on speed
Let's look at the leaderboards first—the gap between the three is quite large.
In Artificial Analysis's latest text-to-image leaderboard, GPT Image 2.6 Sunburst (max) ranks first, Seedream 5.5 ranks tenth, and H3-Image ranks twenty-third.
Image source / Artificial Analysis official website
On the image editing leaderboard, OpenAI still holds first place, ByteDance rises to sixth, and MiniMax climbs to twelfth. Compared with generating images from scratch, the latter two rank somewhat higher in editing capability.
Image source / Artificial Analysis official website
OpenAI, having taken two first places in a row, can it really leave the other two behind in actual use?
We tested the three new models through ChatGPT, Doubao, and Hailuo AI respectively. Among them, ChatGPT used GPT Image 2.6 automatic mode, Doubao called Seedream 5.5, and Hailuo AI tested both the standard and Highspeed modes of H3-Image.
It should be noted that OpenAI's GPT Image 2.6 offers different model tiers through the API, and developers can choose as needed. But most ordinary users generate images directly in ChatGPT, so we did not call the API separately and instead tested the actual experience on each company's product side.
The first task focuses on comprehension.
There's no need to elaborate on how popular short dramas are. First there were fruit-ethics dramas in which pineapples betray strawberries and bananas cheat on grapes, and then there were melodramatic short dramas in which wood ear mushrooms and manicures vie to be heirs to wealthy families. Short-drama protagonists have long ceased to be limited to humans—even food has started dating and fighting over family fortunes.
Recently, the wildly popular extra-long-name tanghulu naturally did not escape this either. With a list of ingredients as long as an incantation, it successfully infiltrated high-society circles. In its family, the more ingredients and the longer the name, the nobler the bloodline—ordinary hawthorn tanghulu does not even qualify to inherit the family fortune.
We asked the three to create a still from a wealthy-family meet-the-parents drama: the heir of a tanghulu family combining more than a dozen ingredients such as milk skin, strawberries, and taro paste brings an ordinary hawthorn tanghulu home to meet the parents. A wealthy young master falls in love with an ordinary girl, the parents disdain the woman's background, and the plot follows the classic mismatched-social-status route.
We also specifically emphasized that all characters have human bodies and tanghulu heads, the ingredients must blend naturally with the facial features, and no ordinary human faces may appear.
More than a dozen ingredients, multiple characters, and a complex setup—one task tests comprehension, execution, and visual expressiveness at the same time.
Prompt (scroll up and down to view the text):
Theme: The extra-long-name tanghulu family, battle over the right to inherit a ten-billion fortune
Generate a cinematic 3D animated still from an absurd Chinese AI short drama, aspect ratio 9:16, exquisite semi-realistic 3D animation style, with bizarre, exaggerated character designs and a strong comedic effect.
Core setup: a top-tier wealthy family made up of extra-long-name internet-famous tanghulu. All characters have normal human bodies and wear luxurious suits and gowns, but their entire heads are giant three-dimensional tanghulu stacked from more than a dozen different dessert ingredients.
Note that these tanghulu are not traditional skewers of a few hawthorns, but are jointly composed of ingredients such as hawthorn, strawberries, Shine Muscat grapes, blueberries, mango, milk skin, taro paste, blood glutinous rice, mochi, cheese, mango pomelo sago, crushed pistachios, chocolate coating, cold-brew yogurt, and butter crisps.
Different ingredients are layered one on top of another: bright red hawthorn and strawberries, emerald green grapes, golden yellow mango, milky white thick milk skin, light purple taro paste, deep purple blood glutinous rice, soft white mochi, dark brown chocolate, and golden butter crisps interweave with one another; some fruits are cut open, revealing filled cheese and yogurt.
The entire head is like a giant, colorful dessert pagoda, clearly extending upward and much taller than a normal human head. The outer layer is wrapped in a crystal-clear sugar coating, with a slender bamboo skewer exposed at the top, and the syrup glistens under warm lighting. Ingredient details are realistically distinguishable, with obvious thickness and three-dimensional structure between layers.
The tanghulu heads have eyes, eyebrows, and mouths naturally integrated into the food shapes, able to express rich emotions such as anger, arrogance, and shock, without appearing as ordinary human faces.
Scene: an extremely luxurious private banquet hall of a wealthy family. A huge golden crystal chandelier, a high ceiling, white marble floors, and floor-to-ceiling glass windows. Outside the window is a bustling city night street with flickering neon lights.
In the center of the banquet hall is a long marble dining table, set with tall red wine glasses, golden candlesticks, and exquisite tableware.
Characters and plot: on the left side of the image is an imposing elderly male head of the family, wearing a dark brown premium suit, a gold ring, and a family crest on his chest. His head is an extremely enormous, super-luxurious tanghulu made of more than twenty layers of different fruits and desserts, with extremely complex ingredients stacked upward layer upon layer, forming an exaggerated crown-like shape. He has an angry expression and extends his right hand, pointing to the other side of the dining table.
Beside him stands a family matriarch, wearing a creamy white silk gown, her head formed by green grapes, milk skin, pistachio cheese, white mochi, green matcha crispy shell, and a golden sugar shell stacked layer upon layer, forming an elegant and complex green-and-cream-white tanghulu tower.
On the right side of the image is a young heir to the wealthy family, wearing a finely tailored white-and-gold suit, standing tall and upright, head raised in confrontation with his elder. His head is a huge colorful tanghulu composed of strawberries, hawthorn cakes, mango pomelo sago cheese, taro paste, blood glutinous rice, mochi, milk skin, pistachio cold-brew yogurt, and chocolate butter crispy flakes, with complex layers and a shape like an exquisite ten-plus-layer dessert tower.
Next to the heir is a young woman, wearing a light green high-end gown, with an ordinary hawthorn tanghulu for a head, and a shocked expression.
Composition and style: use a symmetrical panoramic composition, with the dining table in the center of the image and the four characters arranged on both sides, with all characters fully in frame from head to toe. The tanghulu heads are tall and eye-catching, occupying an obvious visual proportion in the image, but must not obscure the other characters. The characters' body proportions are natural, their clothing is gorgeous and exquisite, and the tanghulu heads are exaggerated yet realistic, creating a strong contrast.
Cinematic 3D animation rendering, exquisite food materials, crystal-clear sugar shells, the crispy texture of milk skin, the dense texture of cheese, the soft texture of mochi, the vivid colors of fruit, and rich highlights and reflections.
First, speed: MiniMax really is fast. However, we almost didn't notice.
At first, we called H3-Image through MiniMax Code and found that generating one image took about 1 minute, and a single edit even took 5 minutes. What happened to the promised 10-second image generation?
Later, we switched to Hailuo AI, and the speed finally picked up. H3-Image in standard mode takes about 25-30 seconds per image, and high-speed mode takes only 10-15 seconds. The reason is that MiniMax Code, as an Agent product, involves task scheduling and other steps in addition to calling the image model, and the whole process incurs extra time.
By comparison, the models from OpenAI and ByteDance both take about 1 minute per image generation. On speed, MiniMax leads.
Now let's look at the results. ChatGPT clearly understands this kind of abstract art. After more than a minute, a still from a wealthy-family short drama was produced. The characters are gorgeous and the expressions are exaggerated, and the tanghulu and facial features are blended quite naturally. Luxurious and dramatic, as if the next second someone will slam the table in anger and throw out a check: "Here are five million, leave my son."
Still generated by ChatGPT
Hailuo AI generates 4 images at a time. The image quality difference between standard mode and high-speed mode is not big, but the mistakes they make are the same. Their first images all had obvious clipping, with characters standing directly on the table, turning a wealthy-family meeting-the-parents scene into the wealthy family collectively "getting on the table."
Image generated by MiniMax H3 standard version
The remaining three images, although without obvious clipping, still didn't get the heads right. Everyone had an ordinary human face, with a large skewer of tanghulu standing on top of their head, as if wearing a deluxe tanghulu hat. Some characters' expressions were also rather flat, lacking a bit of dramatic flair.
We selected the image with the biggest problems as a reference and asked Hailuo AI to put the character standing on the table back down and then blend the human face with the tanghulu. The four regenerated images became a collective "getting on the table," all with clipping problems, and even more characters were on the table, as if performing some mysterious ritual. As for the faces and tanghulu, they still each grew their own way.
Image after modification by MiniMax H3 standard version
Doubao also stumbled on the heads. There were no major problems with the scene or clothing, but the characters likewise had ordinary human faces, and the tanghulu and facial features also each grew their own way.
Still generated by Doubao
Fortunately, it supports local editing. We selected the characters and asked it to change the heads, but out of four faces it only changed three, and the remaining character continued to have a human face, mixed into the tanghulu family.
We had to remind it again. This time we waited about 7 minutes, and finally dealt with the one that slipped through. But after the modification, although the food and human faces barely merged together, the characters' original expressions and identity features became blurred, and the young master and the father looked as if they had swapped identities. This is still far from what Seedream 5.5 claims about "stable multi-round editing"!
Image after modification by Doubao
Next is the second test, which examines aesthetics.
Previously, a Chinese heavenly palace panorama went viral online: the heavenly court above a sea of clouds, with palaces, immortal mountains, and white jade heavenly stairs all present, ethereal and transcendent. This time we set no complex characters; we only asked the three AIs to create a magnificent mythological heavenly court, to see whose aesthetics are more on point.
Prompt (scroll up and down to view the text):
Original Chinese mythological heavenly court panorama, a grand Eastern wonderland on an almost cosmic scale. The camera is positioned above the ninth heaven, using an ultra-wide-angle high-altitude overhead view to present a heavenly city suspended in a boundless sea of clouds. Hundreds of classical Chinese palaces, jeweled towers and jade halls, and vermilion pavilions spread out in layers upon giant floating immortal mountains; the palace gates and watchtowers are distributed in staggered fashion along the mountain contours, extending toward the distant horizon and forming a magnificent yet orderly architectural complex.
At the center of the image is the Lingxiao Treasure Hall, built on a giant white jade platform, with a double-eaved hip roof, golden glazed tiles, giant vermilion columns, gilded bracket sets, and exquisite dragon-and-phoenix patterns carved on the roof ridges. Surrounding the main hall are dozens of immortal palaces of varying sizes; white jade long bridges span the sea of clouds, connecting the floating peaks. Thousands of heavenly steps spiral upward along the mountain ridge, finally converging at the Lingxiao Treasure Hall at the highest point. The roads, stairs, and bridges between the palaces have reasonable connective relationships, and every building has a clear supporting structure and authentic classical Chinese architectural proportions.
On the left side of the image, a giant suspended immortal mountain rises from the clouds; vigorous ancient pines grow on the cliffs, dotted with pavilions and towers roofed in cyan-green glazed tiles. On the right is a grand heavenly palace complex, where layered palace walls, white jade courtyards, and golden flying eaves interweave and extend in the sunlight. Several giant waterfalls pour down from the suspended cliffs, falling into the bottomless sea of clouds below. In the distance, three-tiered immortal mountains loom faintly in the morning mist, and countless tiny palace silhouettes remain at the horizon, creating extremely spectacular spatial depth.
The sky presents a natural gradient of pale cyan, azure, and gold; morning sunlight shines from the east, piercing through the layered sea of clouds and forming clear volumetric light among the immortal palace roofs, white jade long bridges, and mountain ranges. The sea of clouds has a realistic sense of flow, thickness, and natural occlusion relationships; nearby buildings have fine textures, while distant palaces gradually become smaller and paler with distance, forming accurate atmospheric perspective.
It blends the mood and negative space of classical Chinese landscape painting, the majestic composition of Song dynasty landscape painting, the refined details of meticulous heavy-color painting, and the realistic visual effects of top-tier Eastern mythological films. The architecture follows traditional Chinese palace forms; authentic flying eaves, bracket sets, mortise-and-tenon joints, roof-ridge divine beasts, stone-carved balustrades, dragon patterns, and cloud-thunder patterns are clearly visible. The overall palette is dominated by white jade, vermilion, gilded gold, cyan-green, and pale cyan; the colors are bright, transparent, rich, and restrained.
The image is sacred, solemn, serene, and magnificent, with a strong sense of architectural gigantism and epic scale. The emphasis is on highlighting the grand mass of the heavenly palace complex, its spatial order, and the hierarchical relationship between heaven and earth; no people, immortals, heavenly soldiers, or humanoid creatures appear. Cinematic digital concept art, ultra-wide-angle high-altitude overhead view, magnificent deep composition, 16:9 horizontal composition, ultra-high detail, 8K visual texture.
Negative prompt: people, humanoid statues, Western Gothic churches, Greek temples, European castles, European angels, modern cities, glass curtain walls, science-fiction architecture, cyberpunk neon, spaceships, cartoon style, plastic texture, incorrect architectural structure, imbalanced roof proportions, repeatedly duplicated pavilions, unsupported bridges, chaotic suspension relationships, broken stairs, incorrect perspective, overly crowded buildings, heavy fog obscuring the subject, excessive golden light, excessive purple, random text, Logo, watermark, existing film and television characters, game characters, and IP symbols.
ChatGPT continues on the refined route. The scale of the heavenly palace is quite astonishing; palaces densely fill the image, architectural details are maxed out, and the mist-shrouded atmosphere is also in place.
Heavenly palace image generated by ChatGPT
It is just that it seems to have its own obsession with the word "magnificent." The entire heavenly court is resplendent in gold and jade, with so many palaces that they almost cannot fit; it is as if it wants to squeeze every building project of the ninth heaven into one image, and the more you look, the more it feels a bit nouveau riche. I wonder whether, having been No. 1 for so long, it no longer knows how to keep a low profile even when building a heavenly palace.
This is a good opportunity to test its editing ability. We directly pointed out the problems on the image and asked it to reduce the number of buildings, tone down the opulence, and add some Chinese charm. After about 1 minute, the revised version came out: fewer palaces, a cleaner image, and the style was right. Its willingness to take advice: full marks.
Heavenly palace image after ChatGPT modification
The heavenly palace layout of Hailuo AI is relatively more spacious, with clouds and mist weaving among the pavilions, like a finely produced xianxia game scene image. Moreover, whether in standard mode or high-speed mode, the 4 generated images have fairly consistent styles, with the differences mainly in building arrangement.
Heavenly palace image generated by MiniMax H3 standard version
Doubao's performance is very close to it: palaces advance layer by layer, the composition is natural, and it likewise has a strong xianxia flavor.
Heavenly palace image generated by Doubao
After two rounds of testing, OpenAI's model still leads in overall performance: it understands complex instructions most accurately, and its modifications are fast and precise, basically getting it right in one pass, except that its aesthetics occasionally overdo it. MiniMax's model showed a speed advantage, producing images on Hailuo AI in as fast as a dozen or so seconds, and the image quality is not bad either, but when facing complex tasks, it has not yet achieved both speed and quality. ByteDance's model has good image quality, but in this hands-on test, its modifications were both slow and not stable enough, which somewhat disappointed us.
02 Models are getting stronger, and vendors with scenarios are more likely to win
Vendors continue to invest in image model research and development because they have their eyes on the commercialization space of the professional market.
Compared with entertainment needs such as generating avatars and emoji packs, industries such as advertising, e-commerce, and game art need to continuously produce large numbers of images and are also more willing to pay for efficiency. As model capabilities improve, AI has the opportunity to take on more design and asset production work, helping companies save labor costs.
There is more than one way for image models to monetize. In addition to charging individual users membership fees, vendors can also earn revenue through API calls and enterprise services, and can also integrate image capabilities into existing products to increase user usage frequency.
But professional users also have higher demands. Images not only need to look good, but also need to ensure consistency of products, people, and brand elements, and be accurately modified as required. From the tests above, the current models still have quite a few problems to solve.
Facing the same market, the three companies have different strategies.
OpenAI competes on model capability and entry points. ChatGPT has a huge user base; users do not need to open dedicated image-generation software, and can generate and modify images right in the chat box. OpenAI disclosed that users worldwide generate 4 billion images per week on ChatGPT, Codex, and the API. Every time it upgrades its image model, OpenAI can directly reach this group of users without promoting a new product from scratch. Conversely, image capabilities can also increase ChatGPT usage frequency and attract users to pay.
ByteDance, meanwhile, has more mature content creation scenarios.
Seedream can be integrated into products such as Jimeng, Doubao, and Doubao Work, and can also provide services to enterprises and developers through Volcano Ark. Ordinary users can use it to generate fun images, professional creators can produce assets, and enterprises can also generate content in batches through the API.
Image source / Screenshot from the Seedream official website
Jimeng in particular already gathers a large number of users with image and video creation needs, and every Seedream upgrade can directly serve these needs. ByteDance can also collect usage feedback through existing products to continue improving the model.
MiniMax is in a relatively more difficult position. Although it has creation products such as Hailuo AI, its user scale and scenario maturity are far below those of ChatGPT and Jimeng. Even if its model capabilities are good, getting users to switch to a dedicated tool requires additional customer acquisition costs.
Therefore, MiniMax chose to seek breakthroughs in speed, price, and an open strategy.
In this test, the image quality of H3-Image's high-speed mode showed no obvious gap with the standard mode. For users who need to generate images in batches, faster speed means higher efficiency. Coupled with relatively low generation prices, MiniMax has a chance to attract cost-sensitive developers and enterprise customers.
But speed and low prices are unlikely to form a long-term barrier. OpenAI and ByteDance can likewise increase speed and cut prices, and vendors with larger user bases are more likely to dilute costs through economies of scale. What professional users care about more is whether an image can be corrected correctly in one pass as required. If an image has to be revised several rounds, the few dozen seconds saved on a single generation does not mean much.
Opening up model weights has greater long-term value for MiniMax. Enterprises can carry out local deployment and customization according to their own needs, reducing dependence on third-party API services. Compared with closed-source models that can only be called through APIs, this is more attractive to enterprises and developers with customization needs. Once an enterprise completes deployment and system integration around a model, switching models entails additional costs, giving MiniMax a better chance to retain customers.
MiniMax has proven that it can generate images faster and more cheaply, but in the next test, we hope it can not only generate images in a dozen seconds, but also get the heads of the Tanghulu family right in one pass.
This article comes from the WeChat public account "AIX Finance," author: AIX Finance team, published by 36Kr with authorization.