Google has officially launched Gemini, its groundbreaking generative artificial intelligence model. Promoted as their most robust AI model to date, Gemini boasts advanced multimodal reasoning capabilities, catering to three distinct use cases—Ultra, Pro, and Nano—making it adaptable across data centers and mobile devices.
Gemini’s rollout begins with Bard’s AI assistant in two phases. From December 6, a customized version of Gemini Pro will be integrated into Bard, enhancing capabilities such as reasoning, planning, and understanding in English. Early 2024 will see the introduction of an Advanced Bard version, granting users access to more sophisticated models, starting with Gemini Ultra.
Prior to public release, Google rigorously tested Gemini Pro, demonstrating its superiority in six out of eight industry-standard tests when compared to GPT-3.5. The model has been fine-tuned for Bard, showcasing enhanced skills in comprehension, generalization, reasoning, coding, and planning. An illustrative example involves crafting an intricately detailed paper airplane using Bard and Gemini Pro, showcased in a video.
While textual cue support is already available to users, other modalities will soon follow suit. Gemini will debut in English across 170 countries and territories, with plans for expansion into additional languages and regions, particularly in Europe.
Anticipated early next year, Gemini Ultra, the pinnacle of the AI lineup, will be featured in an extended Bard version. Tailored for ultra-complex tasks, this model excels at swiftly comprehending and responding to various forms of information—text, image, audio, video, and code. Security checks for Gemini Ultra in Bard Advanced are underway, with a testing program involving trusted users set to commence soon. The broader launch of Bard Advanced is slated for early next year, heralding a new era in AI capabilities.