A few weeks ago, Google unveiled its new generative AI model, Gemini, which replaces Google Bard. Many talk about its technical capabilities, how easy it is to access its features, but also some bad things.
however, Google has messed up names again., because they also recently presented a second generative AI: Gemma. The name is very similar to Gemini, which can cause obvious confusion among users, because they are not exactly the same. Considering this situation, from Gemini vs. Gemma to determine what is the difference between the two Google AIs.
The main difference: Gemma is an open AI model and Gemini is closed
In case you didn’t know, Google introduced a new open source artificial intelligence model called Gemma. The key feature in this new AI is its biggest difference compared to Gemini, so it’s aptly called “open”.
Unlike Gemini, which is a closed model over which Google has full control, Gemma is an open AI that anyone can download, modify and use as they wish. The same happens with Chromium and AOSP, which are open source versions of Chrome and Android.
Gemma is designed for those companies or people who want to develop themselves thanks to AI, but do not have the necessary resources to develop their own model. And be careful, we are not only talking about money, but also time or technical skills to get it.
Of course, Gema can be downloaded and compiled for use on compatible computers and mobile phones, but at the same time You can use cloud services. Like Google Collab or Hugging Faces to run it.
This AI is pre-trained and uses the same system as Gemini, but the best thing is that it can be adjusted according to the user’s needs. Additionally, Gema has two types of licenses: Free use and commercial use, something that is not available in Gemini.
Gemma comes to compete with Lama-2, an open source AI model from Meta (Facebook). By the way, Google has provided some benchmarks to compare them and Gemma eats its rivals alive in almost all tests.
For end users and developers, another difference between Gemini and Gemma
Now, another key thing that differentiates both generative AI models is who they target. To keep it simple and avoid getting into the swamp, we can break it down as follows: Gemini is aimed at end users. Gemma is aimed at developers..
In fact, unless you do a little digging, there’s a good chance you won’t even hear about it anymore. because of? The new tools developed from this AI don’t take the name, so at most they say something like “Gemma-based”. rather than You always hear about Google Gemini Because it is AI that Google has provided for us.
Finally, there is something you should know: Gemini comes in three different models, while Gemma appears in two. What are these parts? they are –
Gemini: It has Nano, Pro and Ultra versions, from the simplest model to the powerful ordered. Nano is intended to work on devices locally, while others are designed for online support. In addition, two chatbots were born from the latter available to users: Google Gemini (Pro) and Gemini Advance (Ultra). Gemma: Available in two versions, Gemini 2B is the lightest and Gemini 7B is the most complete, with 50% more power. Each model is designed for different needs, although both are available to download and upgrade.
Do you already understand what Gemma is about and how it differs from Gemini? If everything is clear to you, tell us in the comments.
