Hi all,

I just released 4AI version 7.0 and wanted to share why it deserved a full new version number instead of being "just" an improvement over version 6.1. After all, reading the changelog, you would think it only adds support for a few new AI models from Google and OpenAI, not that much of a deal, right?

Well the new image model in 4AI 7.0 makes quite a change: first, the name! Nano Banana, how do they come up with such names? it's in fact called Gemini 2.5 Flash Image but yes, Nano Banana sounds way better.

But what is really better is what this model brings to the table for image creationg and manipulation. We had to rebuild 4AI image tab to get up to what's possible with it, hence the major version change. Let's see what you can do now...

How is 4AI with Nano Banana different?

Nano Banana changes everything about images because it's a mix of an image creator with a "standard" large language model, Google Gemini in this case. So it's understands much better what you ask it, but is also more aware of the context.

As for image modifications, it's now really able to change image without destroying them, and that's progress!

First, here is an example of what context-aware means/ Let's create an image with this prompt:

The Eiffel tower as viewed from the Champs de Mars, with people relaxing on the lawn.

Now let's change the prompt just a little bit:

The Eiffel tower in July 1888 as viewed from the Champs de Mars, with people relaxing on the lawn.

See the difference? The Eiffel tower in July 1888 was about half-completed ( it was ready in march of 1889). I did not have to tell Nano Banana this.

Just mentioning the date is enough for it to change the tower - and people on the ground - appearance. I'm not saying this is an historically accurate image, but it does illustrate what can be done. 

Multi-turn iterations

The other major improvement is the ease and quality with which you can first create an image but also modify it. As Nano Banana is able to apply changes you want to an image without destroying or modifying the entire image, it's now possible to refine an image to what you actually want:

Note: In this video, I sped up a bit the prompts typing but the response time from Nano Banana is real-time, it takes 8 to 12 seconds to get a full-size image.

Modify all your images, not just the AI-generated ones

And now the icing on the cake: what I said about easily modifying images works for any image in your Joomla editor. Not just the ones created with 4AI, but any image.

Start by selecting an existing image, then open 4AI and that's it: you can now tell 4AI things like "Remove the blue car in the background" or "Add some clouds to the sky".

That's all for now, more coming your way soon! 

Cheers,

Yannick