Have you heard of Gemini? It’s Google’s latest generative AI platform.
What makes Gemini special is that it’s ‘natively multimodal’. Here’s what that means… it doesn’t just deal with text; it can handle audio, images, videos, code, and different languages. That’s a big deal, setting it apart from other models that only understand and generate text.
There’s been a little confusion: Gemini is not Bard. Bard is like an app for certain Gemini models. Gemini, on the other hand, is the family of models powering it. To put it in OpenAI terms, Bard is like ChatGPT, and Gemini is like the powerhouse language model (GPT-3.5 or 4).
Gemini has loads of uses, from transcribing speech to captioning images and videos, and even generating artwork. But not all these features are fully baked yet. Google’s promising a lot, but it’s still a work in progress.
Google claims that Gemini outshines the competition in benchmarks, beating GPT-4 in 30 out of 32 widely used academic benchmarks. However, it’s not all sunshine and rainbows. Early impressions suggest Gemini Pro struggles with some tasks and makes factual errors. So, we’ll have to wait and see.
For now, Gemini Pro is free in Bard, AI Studio, and Vertex AI (during the preview phase). But once it exits preview in Vertex AI, it’ll cost £0.0025 per character for input and £0.0005 per character for output. That means a 500-word article might cost you around £5 to summarise and just £0.1 to generate.
It’s only answering text-based queries in English in the U.S. right now but stay tuned for more updates. Are you keen to try it out?
#Google #Gemini #AI
Google Gemini: Everything you need to know about the new generative AI platform
