Google has introduced Gemini, a powerful suite of generative AI models, apps, and services. Developed by Google’s AI research labs DeepMind and Google Research, Gemini comes in three versions: Gemini Ultra, the flagship model; Gemini Pro, a lighter version; and Gemini Nano, a smaller model for mobile devices like the Pixel 8 Pro.
Also read: Sora: OpenAI’s Amazing Video Creator – What You Need to Know
Gemini stands out because it’s “natively multimodal,” meaning it can work with more than just words. Unlike models like Google’s LaMDA, which only deals with text, Gemini models are trained on various data types, including audio, images, videos, codebases, and text in different languages.
The Gemini apps on the web and mobile are an interface to access certain Gemini models. It’s essential to note that Gemini is separate from Google’s Imagen 2, a text-to-image model available in some development tools.
What Can Gemini Do?
Gemini models, being multimodal, have the potential to perform various tasks like transcribing speech, captioning images and videos, and generating artwork. While some capabilities are still in development, Google promises a wide range of functionalities in the near future.
Gemini Ultra:
- Can assist with physics homework, problem-solving, and identifying mistakes in answers.
- Able to extract information from scientific papers and update charts with generated formulas.
- Supports image generation, but this feature is not yet available in the productized version.
- Accessible through Vertex AI and AI Studio, and requires a subscription to the Google One AI Premium Plan priced at $20 per month.
Gemini Pro:
- Improved reasoning, planning, and understanding compared to LaMDA.
- Better at handling longer and more complex reasoning chains than OpenAI’s GPT-3.5.
- Gemini 1.5 Pro can process more data, including up to 11 hours of audio or an hour of video in various languages.
- Available via API in Vertex AI for text input and output, with an additional endpoint for text and imagery.
- Free to use in Gemini apps during the preview period; pricing details for the final version not provided yet.
Also read: Sora’s Big Debut: OpenAI’s New Smart Tool is Now Available
Is Gemini Better than GPT-4?
Google claims Gemini’s superiority on benchmarks, exceeding current state-of-the-art results on 30 out of 32 widely used academic benchmarks. Gemini Pro is said to outperform GPT-3.5 in tasks like summarizing content, brainstorming, and writing. However, some users and academics have raised concerns about Gemini Pro’s accuracy in basic facts, translations, and coding suggestions.
How Much Will Gemini Cost?
Gemini Pro is free during the preview period in Gemini apps, AI Studio, and Vertex AI. Once it exits preview in Vertex, users will be charged $0.0025 per character for input and $0.00005 per character for output. Pricing is per 1,000 characters, and for models like Gemini Pro Vision, per image ($0.0025). Pricing for Gemini Ultra is yet to be announced.
