Google has recently introduced Gemini, its latest and most advanced large language model, marking a significant step in the company’s AI journey. Described by CEO Sundar Pichai as the first in “a new generation of AI models,” Gemini represents a multimillion-dollar investment by Google and is touted as a model with unparalleled flexibility and capabilities.
Developed as an integrated multimodal AI, Gemini is designed to process and reason across various types of inputs. Its training enables it to adeptly handle complex coding requests and solve physics problems, among other tasks. The model’s foundation is likened to a stack of smaller models working in unison, allowing for seamless interaction and understanding of diverse inputs.
Three Sizes of Gemini: Nano, Pro, and Ultra
Gemini is being developed in three versions: Nano, Pro, and Ultra. The Nano version is tailored for on-device use, ensuring accessibility and convenience. Gemini Pro, meanwhile, will be integrated into Google’s improved chatbot, Bard. This enhanced Bard chatbot will be available in the same 170 countries and territories as the current service, expanding its reach and functionality.
In terms of performance, Gemini Pro has already demonstrated its prowess by outscoring GPT-3.5, the model initially powering ChatGPT, in six out of eight AI benchmarks. This achievement is noteworthy, though comparisons with OpenAI’s GPT-4 are still pending.
Gemini Ultra: A Future Benchmark in AI
Gemini Ultra, the most ambitious version of the model, is not expected to be available until at least 2024. In preliminary testing, Gemini Ultra has scored higher than any other model, including GPT-4, on several benchmarks. However, it requires additional testing before being released to a select group of customers, developers, partners, and safety and responsibility experts for further evaluation and feedback.
Also Read
This thorough testing phase underscores Google’s commitment to ensuring that Gemini Ultra is not only highly capable but also safe and responsible in its deployment. The anticipation for Gemini Ultra’s release highlights the growing competition in the AI industry and the continuous push for innovation.
