Image Generation

Meta launched Muse Image, a new proprietary AI model for image generation, signaling a strategic pivot to cultivate creators and advertisers and unlock new revenue streams. Accessible for free via Meta AI apps and messaging, with premium options for power users, Muse Image will also power advertiser tools within Advantage Plus, streamlining ad creative development. This move diversifies Meta’s revenue beyond traditional advertising, with future plans for an AI video generation model, Muse Video.

Google has launched Nano Banana Pro, an AI image editing and generation tool powered by Gemini 3 Pro. The new capabilities include infographic creation, character consistency across multiple images, and visualization of non-visual data. The tool is available in the Gemini app with limited free quotas, and wider access for subscribers. Google is also introducing a feature to verify AI-generated images, addressing misinformation concerns, as competition with OpenAI’s ChatGPT intensifies. Google aims to scale its AI offerings, exploring potential in creative and other industries.

Google’s AI Mode is evolving to include visual search results, integrating image generation capabilities. This expansion, powered by Gemini 2.5 and other Google technologies, addresses queries better suited for visual answers, like shopping inspiration and interior design ideas. Users can refine results with follow-up prompts, tailoring visuals to their preferences. This move, driven by competition from AI alternatives, aims to enhance user experience and maintain Google’s leadership in the AI-driven search landscape, potentially impacting e-commerce and online advertising.

Nano-banana, a Google AI Studio image model accessible through Gemini, has emerged as a top performer in the LMArena, demonstrating remarkable consistency and prompt accuracy. It excels at tasks like image refinement, facial perspective manipulation, and composite creation while maintaining subject identity. Its strength lies in its deep image understanding, enabling repairs, style alterations, and even simple 3D model creation. However, nano-banana exhibits limitations in style changes and localized editing, and requires explicit prompts due to strict safety protocols.