Google’s Nano Banana 2 AI model reportedly made a brief appearance on the Media AI platform.
Nano Banana 2 is said to offer better prompt understanding and instruction following
Photo Credit: Google
Google reportedly previewed the Nano Banana 2 artificial intelligence (AI) model a couple of days ago. As per the report, the upcoming model comes with several improvements across prompt understanding, instruction following, and visual quality. However, it is not the only AI project the Mountain View-based tech giant has currently undertaken. The company is currently working on at least five different projects that have become public knowledge via various leaks and reports. From the Gemini 3 Pro AI model to the web-browsing agent Jarvis, let us take a look at the upcoming Google AI projects.
According to TestingCatalog, Google's Nano Banana 2 became briefly available on the Media AI platform. The new version of the company's popular image editing AI model is said to offer improved capability in colouring precision, the ability to process complex tasks, higher instruction following, and correction of textual elements in an image.
Interestingly, based on internal observations of Nano Banana 2, the report claims that Google is not shifting the AI tool to Gemini 3.0 Pro, and instead sticking to Gemini 2.5 Flash. What that means is there will not be any transformative inclusions, but it also means that users will have to wait a lesser time before it becomes available. So far, the tech giant has not revealed any release date for the same.
After Google CEO Sundar Pichai confirmed during the Q3 2025 earnings call that the company is trying to bring the Gemini 3 models before the end of the year, several leaks have highlighted what it might offer. Several leaks have claimed that two mystery models, lithiumflow and orionmist, that appeared on the crowd-sourced AI model ranking platform LMArena, could be the company's attempt at testing the model in stealth.
These models were said to offer improved structural reasoning, which is visible in their front-end code generation of complex and interactive visualisation using SVG and HTML. It is said to also demonstrate a marked improvement in mathematical reasoning, solving complex queries that previously presented significant difficulty for Gemini 2.5 Pro.
In October 2024, a report claimed that the tech giant is working on Project Jarvis. It was described as an AI agent that will let users automate tasks such as purchasing a product from an e-commerce website, booking tickets, filling out forms, and more. However, its functionality is said to be limited to browser usage.
Later, in November 2024, another report claimed that Jarvis made a brief appearance on the Chrome Web Store. The listing page is said to have described Jarvis as “a helpful companion that surfs the web with you.” It was rumoured to be released last year, but has not been seen so far.
Google is also rumoured to be planning new AI capabilities for its next flagship smartphone, the Pixel 11 series. One of the leaked features is described as “on-device Night Sight Video,” which can enhance video quality in low-light conditions immediately after capturing it. The company is said to be using a custom on-device AI model to power the real-time processing.
Last month, the tech giant published a blog post announcing a new capability in the Gemini sidebar in Workspace apps such as Google Drive, Docs, Sheets, and more. The AI chatbot was upgraded with multi-step task completion with a single prompt. To further integrate the AI experience for enterprises, the company is said to be working on general-purpose AI agents that can not only handle multiple tasks but can also complete more complex tasks. Currently, Google's multi-step functionality is available in Sheets.
Get your daily dose of tech news, reviews, and insights, in under 80 characters on Gadgets 360 Turbo. Connect with fellow tech lovers on our Forum. Follow us on X, Facebook, WhatsApp, Threads and Google News for instant updates. Catch all the action on our YouTube channel.