--- title: "Google’s New AI Model Works Offline on Phones Using Just 2GB of Memory" description: "Google's new memory-efficient AI model runs complex tasks offline on smartphones using only 2GB of RAM, enhancing privacy and accessibility." author: "Meet Patel" published: 2025-06-27 updated: 2025-06-27 canonical: https://www.mindstick.com/news/4603/google-s-new-ai-model-works-offline-on-phones-using-just-2gb-of-memory category: "google" tags: ["artificial intelligence"] reading_time: 2 minutes --- # Google’s New AI Model Works Offline on Phones Using Just 2GB of Memory [Google](https://www.mindstick.com/blog/304082/what-are-the-latest-google-updates-2024) has come up with a new [AI](https://www.mindstick.com/blog/304401/ai-generated-can-be-indexed-on-google) model that is able to make complex functions completely offline on smartphones. This model only needs 2GB of [device memory](https://www.mindstick.com/forum/33525/pick-image-from-device-memory-using-uiimagepickercontroller) and this means that it can be applicable to a variety of the current mid and previous generation phones. It removes the reliance on cloud connection on fundamental AI tasks. ## Highlights: - Google's new AI model runs complex tasks offline on [smartphones](https://yourviews.mindstick.com/view/85155/the-latest-trends-in-smartphone-technology-foldable-screens-5g-more). - It operates using just 2GB of device RAM. - The technology functions without cloud connectivity. - It brings advanced [AI capabilities](https://answers.mindstick.com/qa/114965/how-will-advancements-in-mobile-processor-ai-capabilities-change-everyday-app-functionality) to mid-range and older devices. - Local processing enhances [data privacy](https://www.mindstick.com/blog/304706/google-s-new-data-privacy-technology-everything-you-must-know) by keeping information on-device. The technical accomplishment breaks the memory-intensive nature of advanced AI. Amazingly, Google built the model with new neural structures and optimisation strategies that are designed to run in a constrained mobile environment. This enables real time actions such as [language translation](https://answers.mindstick.com/qa/102687/how-does-google-s-cloud-translation-api-support-language-translation-in-applications) or image analysis to be performed directly on the phone. Local processing guarantees that data on sensitive issues like conversations or pictures are retained at the device. Such an on-device solution offers a considerable privacy and [security](https://www.mindstick.com/articles/337508/enhancing-application-security-with-oauth-2-0) improvement over the cloud-reliant AI systems, where the data is not exported. The simple memory requirement of the model has broad access to sophisticated AI capabilities. It will allow features to its flagship devices, such as more powerful [voice assistants](https://www.mindstick.com/blog/306524/the-evolution-of-smart-homes-beyond-voice-assistants) or intelligent functions within the camera on lower-end smartphones without costly hardware solutions, which will reach a much wider majority of the [global population](https://answers.mindstick.com/qa/108362/what-are-the-recent-trends-in-global-population-growth). --- Original Source: https://www.mindstick.com/news/4603/google-s-new-ai-model-works-offline-on-phones-using-just-2gb-of-memory Copyright © MindStick Software Pvt. Ltd. This Markdown version is provided for developers, AI systems, and offline reading.