Google has unveiled its latest AI model, Gemini, which has surpassed OpenAI’s GPT-4 in key benchmarks, marking a major step in the AI race. Announced on December 6, 2023, Gemini is a multi-modal model that proficiently processes text, images, audio, and video. The flagship Gemini Ultra achieved a remarkable 90.0% on the MMLU language understanding test, eclipsing the human expert benchmark and GPT-4. Strong performance in reasoning, math, and coding tasks highlights its advanced capabilities. Google is launching Gemini in three tiers: Ultra for complex tasks, Pro for integration across its services, and Nano for on-device use. This strategic rollout, starting December 13, boosts accessibility for developers and businesses alike. Gemini’s native multi-modality provisions a seamless data processing experience, affirming Google’s position as a leader in AI innovation. With this competition, consumers stand to benefit from faster advancements and accessible pricing in AI technology.
Source link
