Product Update: More Usage For All, Advanced Voice Mode, New Flux & Stable Diffusion, and More!

Written by:

ai generated image of a robot standing in the middle of a well-lit modern office space with confetti falling from the ceiling

We’ve been working really hard to bring big updates to Magai. There’s a lot packed into today’s update, so let’s get started, shall we?

New Chat Models:

  • Voice GPT
  • Claude 3.5 Haiku
  • Meta Llama 3.2 11B

New Image Models:

  • Flux 1.1 Pro Ultra
  • Stable Diffusion 3.5
  • Stable Sketch
  • Stable Structure
  • Stable Style

New or Improved Features:

  • Re-added image uploader in Editor
  • Redesigned Model Selection Popups
  • See Original Image on upscales
  • Prompt Enhance upgrade

A total of 47 development tickets were completed with this update including various bug fixes and quality of life improvements. Now let’s dig in to some of the juicy details.

New Chat Models

The obvious big upgrade here is the introduction of GPT Voice. This is the equivalent of “Advanced Voice” mode in ChatGPT. You can now speak, in real time, with an AI model who will speak back to you.

I’m excited to hear all the new ideas you come up with to use GPT Voice mode.

We have also added the Claude 3.5 Haiku from Anthropic, and Llama 3.2 11B from Meta. Both models provide exceptional capabilities for low cost at fast speeds.

New Image Models

It seems like these image generators are coming out with new versions of their flagship models left and right. Specifically Flux (my current favorite), and Stable Diffusion.

Flux 1.1 Pro Ultra

The next iteration of Flux’s Pro model is 1.1 Pro Ultra. According to the people behind the model, it’s a high-resolution model that is significantly faster than other models in its class. It produces a much higher resolution image in less time, without compromising quality.

Stable Diffusion 3.5 (Large)

According to Stability AI:

“At 8.1 billion parameters, with superior quality and prompt adherence, this base model is the most powerful in the Stable Diffusion family. This model is ideal for professional use cases at 1 megapixel resolution.”

The next three models on this list are what’s known as “Control” models. These are created to take an input image and allow you to modify the image in different ways.

Stable Sketch

Turn your simple sketches into polished designs! This tool is perfect when you need to explore different design ideas quickly. Just upload your hand-drawn sketch, and watch it transform into a professional-looking image. Working with existing images? No problem – you can easily adjust and enhance them by controlling their basic shapes and outlines. It’s like having a design assistant that takes your rough ideas and makes them look amazing!

Stable Structure

Keep the shape, change the style! This tool is perfect for transforming images while maintaining their basic structure. Want to turn a photo into artwork while keeping everything in the right place? Or recreate a character in a different style? Stable Structure ensures your new image stays true to the original layout while allowing creative modifications.

Stable Style

Copy any style, create something new! Upload an image you love – maybe it’s a painting, photo, or digital artwork – and Stable Style will learn its unique look. Then, just describe what you want to create, and the tool will generate a brand new image using that same style. It’s like having an artist who can perfectly mimic any style you show them!

Feature Additions and Improvements

Although it may seem like the app gets updated every couple of weeks, we strive to make Magai better every single day. Here are just a handful of the things we did over the past couple of weeks.

Image Upload is Back

With our new Editor layout, we accidentally hid the “Upload Image” feature. Shoutout to Steve Dirks for pointing this out.

We’ve brought this feature back–just click on the context menu button and you’ll see “Upload Image” as an option.

Redesigned Model Selection

One of our objectives this cycle was to minimize confusion and make selecting a model more intuitive. And I think we’ve done a good job of that.

The new popup when you select a model includes more obvious filtering, and the ability to search quickly through the list.

The Editor page now has a popup style model selector as well.

Go to Original Image

Previously, when you created a variation of an image through upscaling or inpainting, there was no easy way to get back to the original image. Now we’ve added a button that will allow you to that.

Prompt Enhance Update

When you use the Prompt Enhance button inside the Editor page, you now have the options to:

  • Regenerate the prompt
  • Revert the prompt
  • Clear the prompt

The “Revert” option is the newest addition, allowing you to quickly bring back the previous version of your prompt (just in case you didn’t like the enhanced version).

One More Thing

I announced this to everyone in our Facebook group a few days ago, but it’s worth noting here:

All plans just got a huge increase in monthly usage!

You can now get more AI content out of Magai than ever before. And with new models like Nemotron at such low costs, the possibilities are exponential.

And that’s all for this update. Have a great weekend!

Latest Articles