Google has announced its latest AI model, Gemini, promising a major leap forward in the field of artificial intelligence.
Gemini AI surpasses existing models, including ChatGPT with its multimodal capabilities. These capabilities allow it to process and understand information across various formats like text, code, images, and audio.
The new model comes in three versions, each catering to different needs:
- Gemini Ultra: This most powerful version is designed for highly complex tasks requiring advanced reasoning and problem-solving skills. It boasts a 90% score on the MMLU benchmark, outperforming human experts in tasks encompassing diverse subjects like math, physics, history, law, and medicine.
- Gemini Pro: Ideal for a wider range of tasks, this version offers a balance between power and efficiency, making it suitable for real-world applications in various industries.
- Gemini Nano: Designed for on-device processing, this version brings AI capabilities directly to smartphones and other edge devices. It will initially power features like Summarise in the Recorder app on the Pixel 8 Pro and Smart Reply via Gboard.
While pricing for Gemini’s versions hasn’t been announced, Google plans to offer access through its Cloud platform, similar to how it offers PaLM 2.
This tiered approach ensures accessibility for researchers and developers of all sizes, fostering further innovation in the AI space.
Compared to its rivals, Gemini stands out in several key areas:
- Multimodal capabilities: Unlike models like ChatGPT, Gemini can process and understand information across diverse formats, allowing it to tackle a wider range of tasks.
- State-of-the-art performance: Gemini’s benchmark scores surpass those of existing models, demonstrating its superior ability in areas like language understanding and problem-solving.
- Scalability: With three versions available, Gemini caters to diverse requirements, from research and development to everyday applications on smartphones.
While ChatGPT remains a strong contender in the chatbot domain, Gemini’s multimodal capabilities and superior performance position it as a leader in the broader field of AI.
Here are the technical scores of the Text-based model:
Also, here are the scores for the multimodal:
With its potential to affect various industries, from healthcare and education to customer service and content creation, Gemini marks a major milestone in the evolution of artificial intelligence.
In our previous report, we shared how Google is delaying the launch of its new AI model Gemini. And, because of having issues with other languages apart from English, the launch of the public models might be in 2024.
