Image Generation Model

On October 22nd, BlackForest Labs Local Time released the FLUX 3 Image Klein Image Generation Model on October 15th. FLUX 3 Image Klein has 5.6B parameters, which are optimized for consumption grade GPUs. Its BF16 weight requires 12.2GB memory and FP8 weight requires 5.6GB memory. The INT8 version of this model can be used through Comfy..

On April 22, OpenAI released yesterday (21 April) the image generation model GPT Image 2 was officially fully distributed yesterday and is open to all ChatGPT users, including free accounts. AI raw drawing tool ChatGPT Images 2.0, based on GPT Image 2 model, with core upgrades in the introduction of “thinking ability”. GPT Image 2 has made a huge breakthrough in Chinese-language translation. According to APPSO testing of the model, GPT..

On February 27, according to the news, Google officially launched a new generation of image generation models, Nano Banana 2 (Gemini 3.1 Flash Image), which is based on high-speed generation to further improve the quality, understanding and coherence of drawings, and position them as a light-weight alternative to the Nano Banana Pro, open to a wider range of users. Nano Banana 2 Continue the Gemini series of real-world knowledge banks and search real-time information in conjunction with web pages to make models understand, compare..

On January 14, according to the news, Union Hua was declared today as an open-source new generation image generation model, GMM-Image, based on the roll-out of Atlas 800T A2 equipment and the MindSpore AI framework to complete the full process from data to training, the first SOTA multi-modulate model to complete full training on a national chip. GLM-Image combines image generation with language models using an autonomous and innovative "self-return + proliferation decoder" hybrid structure. 1AI with GLM-Image core highlights as follows..

On November 21, Google officially launched a new image generation and editing model, Nano Banana Pro (Gemini 3 Pro Image). The model, based on the Gemini 3 Pro architecture, is called “unprecedented control, perfect text replicating and enhanced world knowledge reserves” and transforms user ideas into “studio-equality” designs. This time, it's more..

On November 6th, according to the media TestingCatalog, a new up-to-date alert card appeared in the previous Google Gemini web interface. According to its analysis, Google is preparing a new image generation model for the nano Banana series. It was described that the model would be named "GEMPIX2" and published as a second generation of nano Banana. The report analysis, the advance card, is usually a proximate signal, and, according to previous assumptions, such updates usually appear in public..

On November 5, Microsoft launched its first self-study AI image generation model, MAI-Image-1, in October of this year, and today announced its application to Bing Image Creator and Copilot Auto Express (with the temporary exception of the EU). Mustafa Suleyman, Head of Microsoft AI Department, described the model as “excellent” in producing images of food, natural landscapes, artistic light and factual details. Microsoft watch..

Early this morning, Google officially presented its most advanced image generation and editing model - Gemini 2.5 Flash Image (aka nano banana). According to the official introduction, the main features of Gemini 2.5 Flash Image include the following: Fully maintain the consistency of the character: it can easily place the same character in different environments, or show the same product from multiple angles, while perfectly keeping its core subject unchanged. Cue-based image editing: Allows the user to use simple natural language...

April 15 news, fast hand today held "inspiration comes true" can Ling AI 2.0 model conference, and announced that the base model upgraded again, for the world officially released can Ling 2.0 video generation model and can figure 2.0 image generation model. According to the introduction, Ke Ling 2.0 model maintains global leadership in dynamic quality, semantic response, screen aesthetics and other dimensions; Ke Tu 2.0 model significantly improves in command compliance, movie texture and artistic style performance. Gai Kun, senior vice president of Racer and head of the community science line, revealed that since its release in June last year, Ke Ling AI...

Today we share a very interesting AI tool: OmniGen A. What is OmniGen OmniGen is a "unified image generation model", without the need to install ControlNet, IP-Adapter, Reference-Net and other plug-ins, according to the text prompts to automatically recognize features in the input image (eg, an object, pose, mapping). OmniGen is a "Unified Image Generation Model" that automatically recognizes features (e.g., an object, a pose, a mapping) in an input image without the need to install ControlNet, IP-Adapter, Reference-Net, etc. What's the use? For example, in the following two images, if you want to put the characters in them in the same background, it used to be more troublesome, but now it only takes one line of instruction. Ta-da! Used above...

Dec. 8 (Bloomberg) -- Musk's X has introduced a new AI image generation model, "Aurora," for its AI chatbot Grok, which is capable of generating more realistic images than Grok's other image-generation models and appears to have fewer limitations on what can be generated. Like Grok, anyone can use Aurora, which is available in the Grok model selector under the "Grok 2 + Aurora beta" option, though free users will only have limited access. Tech...

Today, we introduce OmniGen, a unified image generation model proposed by the Beijing Institute of Artificial Intelligence, which can be used to perform various tasks, including but not limited to text-to-image generation, subject-driven generation, identity preservation generation, image editing, and image conditional generation.OmniGen does not require any additional plug-ins or operations, and it can automatically recognize features (e.g., desired objects, human poses, depth mapping, etc.) from the input image based on textual prompts. features (e.g., desired object, body pose, depth mapping) based on textual prompts. Related Links Paper: https://arxiv.org/pdf/2409.11340 Code: htt...