Google is upgrading its Gemini chatbot with a new AI image model that gives users a fine control over editing photos, a step means to catch OpenaiI’s popular image tools and attract users from Chatgpt.
The update called Gemini 2.5 Flash Image, starts all users in the Gemini app, as well as rolling out developers through Gemini API, Google AI Studio and Vertex AI platforms.
Gemini’s new AI image model is designed to edit more accurately for images – based on natural language requests from users – preserving the stability of face, animals, and other details, something that struggles the most rivals. For example, a photo may include a chatgate or zai grooc to change the color of someone’s shirt, and the result may include a distorted face or a converted background.
Google’s new tool has already attracted attention. In recent weeks, social media users banged up An impressive AI image editor in Crowdsored Evolution Platform, LMARENA. The model users appeared anonymously under the anonymous name “Nano-Banaa”.
Google says it is behind the model (if it was not obvious already From all bananas related signs), which actually have native image capacity within its major Gemini 2.5 flash AI model. Google says the image model is the state -of -the -art on LMARENA and other benchmarks.
“We are actually carrying forward visual quality, as well as advancing the ability of the model to follow the instructions,” said a product lead Nicole Bichtova on the visual generation model in Google Deepmind in an interview with Techcrunch.
“This update does a better job that edits more basically, and the output of the model is usable for whatever you want to use for them,” Bichtova said.
AI image model has become an important battleground for Big Tech. When Openai launched the original image generator of GPT-4O in March, it removed the use of chatgate through the roof for a frenzy of AI-Janit Studio Ghibali Mems, which according to Openaii CEO Sam Altman, left the GPU of the company “Melting,
To live with Openai and Google, Meta announced last week that it would license the AI image model from the Startup Midzorney. Meanwhile, the A16Z -backed German Unicorn Black Forest Labs dominates the benchmark with its flux AI image model.
Perhaps Gemini’s impressive AI image editor can help Google close its user difference with Openai. Chatgpt now logs to more than 700 million weekly users. On Google’s earnings call in July, Tech Giant CEO Sundar Pichai revealed that there were 450 million in Gemini Monthly User – Applying weekly users is even less.
Brichtova says that Google specifically designed the image model keeping in mind the matters of consumer use, such as helping users imagine their home and garden projects. “World knowledge” in the model is also better and can add multiple references to the same sign; For example, merging an image of a couch, a living room photo and a color palette into a harmonious render.
While Mithun’s new AI image makes it easier for generator users to create and edit realistic images, the company has security measures that can make users. Google has struggled with AI image generator safety measures in the past. At one point, the company apologized for Gemini and generated historically wrong pictures of people, and completely rolled back the AI image generator.
Now, Google feels that this is a better balance.
“We want to give users creative control so that they can get from models that they want,” Bichtova said. “But it is not that anything happens.”
The general AI section of Google’s service conditions restricts users from generating “non-conscience intimate imagination”. The same safety measures do not seem to be present for the grouke, allowing users to make AI-Janit Clear picture Taylor meets celebrities like Swift.
To address the rise of deepfek imagery, which can make it difficult for users to learn what the real online is, Brichtova says that Google AI-made images, as well as the visual watermark to the identifiers in its metadata, applies visual watermarks. However, no person scrolling an image on social media can seek such identifiers.
