Google Unveils Gemini AI Family, But Delays Flagship Model
Google introduces Gemini as a tiered family of AI models rather than a single product, launching the lightweight Gemini Pro while delaying the flagship Ultra version until next year. The initial rollout focuses on powering Bard and select developer tools.
Google officially unveils Gemini, its highly anticipated next-generation generative AI, but the release looks different than expected. Instead of a single model, Gemini is a family of AI systems divided into three tiers: the flagship Gemini Ultra, the mid-tier Gemini Pro, and the mobile-optimized Gemini Nano. The most powerful version, Gemini Ultra, does not launch this week and remains delayed until sometime next year.
The version currently rolling out, Gemini Pro, acts as a lightweight offshoot that immediately powers Google's Bard chatbot for English users in the U.S. Google claims this update brings improved reasoning, planning, and understanding capabilities to Bard, though independent testing remains impossible since the company did not allow press access or live demonstrations prior to the announcement. Enterprise customers gain access to Gemini Pro through Vertex AI starting December 13.
Meanwhile, Gemini Nano targets on-device processing and comes in two parameter sizes to accommodate different mobile hardware capabilities. This smallest model launches in preview exclusively on the Pixel 8 Pro through Google's AI Core app for Android 14, with broader Android developer access opening soon. Over the coming months, Google also plans to integrate Gemini across its wider product ecosystem, including Duet AI, Chrome, Ads, and Search Generative Experience.