Google's Most Powerful AI Model Gemini 1.0 Released, Far Ahead of GPT-4.0
-
On December 7th, news broke that Google first revealed its AI model Gemini under development at the developer conference in May. After 7 months, Gemini has finally arrived.
According to Google's official announcement, Google has officially released Gemini 1.0, the most powerful, versatile, and flexible model built by Google to date.
It is reported that Google has released three different versions for different scenarios:
Gemini Ultra: Google's largest and most powerful model, designed for highly complex tasks.
Gemini Pro: The optimal model for a wide range of tasks.
Gemini Nano: The most efficient model for devices like smartphones.
Google has also compared Gemini with OpenAI's GPT-4, with Gemini outperforming GPT-4 in all benchmarks except the HellaSwag dataset.
Officials stated that from natural images, audio, and video understanding to mathematical reasoning, Gemini Ultra's performance exceeded the current state-of-the-art in 30 out of 32 academic benchmarks widely used in large language model (LLM) research and development.
Gemini Ultra scored 90.0%, becoming the first model to surpass human experts in the MMLU (Massive Multitask Language Understanding) test. MMLU comprehensively uses 57 subjects, including mathematics, physics, history, law, medicine, and ethics, to test world knowledge and problem-solving abilities.
In the new MMMU benchmark, Gemini Ultra also achieved a score of 59.4%. This benchmark consists of multimodal tasks spanning different domains that require careful reasoning.
It is reported that Gemini 1.0 possesses sophisticated multimodal reasoning capabilities, enabling it to comprehend complex written and visual information while identifying obscure knowledge within vast datasets.
Google states that Gemini 1.0 has been trained to simultaneously recognize and understand text, images, audio, and more. This allows it to better grasp nuanced information, answer questions related to complex topics, and particularly excel at explaining reasoning in challenging subjects like mathematics and physics.
In terms of coding capabilities, the first-generation Gemini can understand, interpret, and generate high-quality code in the world's most popular programming languages (such as Python, Java, C, and Go).
According to Google, Bard will use a fine-tuned version of Gemini Pro for more advanced reasoning, planning, and understanding, available in English in over 170 countries and regions. Plans are in place to expand different modalities in the future and support new languages and regions.
Additionally, Google will integrate Gemini into Pixel devices, with the Pixel 8 Pro being the first smartphone to feature Gemini Nano. This will enable new features such as summarization in the Recorder app and introduce Smart Reply in Gboard.
In the coming months, Gemini will be applied to Google Search, Ads, Chrome, and Duet AI.
Starting from December 13, developers and enterprise customers can access Gemini Pro through the Gemini API in Google AI Studio or Google Cloud Vertex AI.
Regarding Gemini, Google stated that this represents an important milestone in the development of AI and also marks the beginning of a new era for Google.