Bard will be accessible in English in more than 170 countries and territories, including India, thanks to its new large language model, Gemini.
Users on social media started to contest Google’s assertions that its new high-performance AI model, Gemini, was far better than OpenAI’s GPT-4.
With the release of Gemini, which it bills as its largest and most powerful AI model, Google astounded the entire globe. Google is joining the fight with Gemini, joining forces with OpenAI, Meta, and Anthropic, whose AI models have been demonstrating significant advancements in their capabilities. Google has released the three Gemini models: Nano, Pro, and Ultra.
The Nano is intended to be included in Google Pixel 8 Pro, even if the Pro version will be incorporated with Google Bard. With so many options available to customers, Google may pose a challenge to OpenAI’s ChatGPT hegemony. Additionally, Gemini is multimodal and effective with a wide range of information, including code, photos, videos, and text.
Currently, ChatGPT vs. Bard
Google’s blog post compared Gemini’s models to other well-known AI models, such as Hugging Face’s Llama-2, Musk’s Grok, and Google’s earlier benchmark model PaLM-2, in addition to the GPT models.
In the Multi-Task Language Understanding benchmark test, Gemini Pro achieved 79.13% as opposed to GPT-3.5’s 70%. Gemini Pro scored a significantly higher 86.5% in the GSM8K benchmark exam, which evaluates arithmetic thinking, than GPT-3.5’s 57.1%. Gemini Pro outperformed GPT-3.5 with 67.7% at the HumanEval benchmarks, which are used to test code creation. GPT-3.5 scored 48.1%. GPT-3.5 performed higher than Gemini Pro in only one benchmark: MATH, where Gemini Pro scored 32.6% as opposed to GPT-3.5’s 34.1%.
Google Gemini Ultra vs. GPT-4
Meanwhile, Google reports that on 30 out of 32 benchmark tests, including reasoning and picture recognition, the unreleased Gemini Ultra fared better than cutting-edge versions like GPT-4.
Bard will continue to develop into what Google refers to as “Bard Advanced” when Ultra is released.
In Massive Multitask Language Understanding (MMLU), Gemini Ultra scored an astounding 90%, demonstrating its comprehension of 57 areas, spanning STEM and the humanities. Conversely, GPT-4V had an 86.4%.
Gemini Pro does better than GPT-3.5.
Millions of people will have free access to Gemini Pro since it will be included with Google Apps and the chatbot Bard. However, the free edition of ChatGPT only provides GPT-3.5. GPT-3.5 was outperformed by the Pro version on six of the eight benchmarks. It is now the most potent free AI chatbot available anywhere in the world. With multimodal features and access to millions, Gemini Pro is likely to surpass ChatGPT. It will take some time before we can fully assess its potential.
