Transcription
Just a couple of days ago, we explored some potential new GPT checkpoints, possibly being GPT 5.1 or even early GPT6 builds being quietly tested on LM Arena's webdev arena as well as the design arena. At the time, we saw four incredibly powerful models focused on UI and UX generation. They're capable of building entire front-end layouts and prototypes with near human design intuition.
Now there's a new wave of experimental variants. Cicada, Caterpillar, Cascilus, as well as Firefly. All four insanely fast designed optimized models that push creative reasoning to a new next level. Now among them, Cicada stands out to be the strongest performer so far and is producing landing pages with stunning precision and aesthetic depth. Just take a look at this cicada generated landing page which looks absolutely amazing with its polished UI and UX design and you can see that it is something that looks production ready straight out of a single prompt. This is something that is even better than the previous GBT checkpoints that we had taken a look at in the last couple of days.
And like I had mentioned, you have the ability to access it through Design Arena, which is a crowdsource benchmarking platform where users can vote on AI generated UIs, prototypes, and front-end designs and head-to-head battles with other models apart of different providers. And you can basically vote on which design is better. And essentially different model providers use these different platforms like LM Marina's WebDev Arena to test out how their model performs up against other model providers to see if it's performing on par with the community's standard. And just to clarify, you have the ability to access it through Design Arena as well as through WebDev Arena completely for free without you paying or creating an account.
Before we get started, I just want to mention that you should definitely go ahead and subscribe to the World of AI newsletter. I'm constantly posting different newsletters on a weekly basis. So this is where you can easily get up-to-date knowledge about what is happening in the AI space. So definitely go ahead and subscribe as this is completely for free.
The previous models are still available through web dev arena as well as on design arena. And you can see that this is a generation of a news aggregator by the willow model. And it looks pretty decent in my opinion in terms of its UX design. But essentially, these four new variants are designed to optimize GPT5 variants, even possibly being GPT 5.1 or maybe even the GPT6 build. Now, these models are pushing the limits in terms of its speed, structure, and creativity in any sort of code generation.
Cicadia is a fast, high precision, and visually striking model. It leads in layout, quality, as well as balance. Caterpillar is a stable, iterative, excels in structured workflows and refinement. Crystis is something that is imaginative and expressive with its code generation. It's great for bold concept designed UI tasks. And Firefly is the lightweight as well as dynamic model. It's optimized for instant rendering and quick prototypes.
Now, overall, all of these tests are showcasing that Cicada is probably the best model. Three of the models have a reasoning budget associated with them. Firefly has zero, Chrysis has 16, Cicada has 64, and Caterpillar is something that is adaptive at the moment. And all of them are showcasing a clear progression in terms of instant generation to deep structured design reasoning, which is used depending on the prompt. In essence, all four of these checkpoints are better than the previous four checkpoints that we had showcased as they are extremely dominating in both speed as well as code generation.
So this is a new test that I have been basically asking every sort of model and that is the prompt to create a butterfly in SVG code that's animated. Now we all know that most models nowadays can create a butterfly in SVG which was a prompt that was pretty hard back in the day. But now the new prompt to test these models to see how well it is in terms of its coding performance is adding the animated part within the prompt which is going to showcase how well the model is in terms of his SVG code capability. And in this case the Firefly model which is probably the most lightweight model that's a part of the new checkpoint series. And in this case Firefly was capable of generating this elegant animated SVG Butterfly. If we actually full screen this, you can see that it is animated, but it's not accurately symmetrical to a proper butterfly wing. So, in that case, it didn't do the best job. But in terms of other capabilities like adding random wing colors as well as its creative animations to the pause button, it did a great job in that front. So, overall, I guess this is a decent generation. It's not perfect it, but it is definitely surpassing all these other generations we can see.
And in my opinion, you have a higher probability of getting a generation by the checkpoints through design arena. So, it is something that gets you a better chance of getting one of the generations out of these checkpoints. And it's super simple. You don't need to create an account. You can simply paste in your prompt like creating a CRM dashboard and within a couple minutes, it will generate it for you. And afterwards, you can vote on which generation is better.
Now, based off of what I see already, this does not look like any of the checkpoints. So, let's just simply click through and see if we got one of the generations. This might look like something that was generated from OpenAI. So, let's prefer to vote on this one. And I believe this one also looks better than this generation. And then after you have finished voting, you're going to be able to see which model generated which response. So, actually, surprisingly, it was not generated by OpenAI. This was something by the DeepSeek version 3 model which looks pretty good in my opinion. But that's essentially how you can use Design Arena.
Finally, I got it so that I was able to create a CRM dashboard with one of the new checkpoints from Firefly. In this case, it looks actually really beautiful in terms of its UI and UX design. This is something that it fully coded out in a single shot. It took approximately 3 minutes to generate, which is pretty decent. And something I forgot to mention is that within Design Arena, you have the ability to export the code itself. In this case, I believe it wrote over a,000 lines of code, approximately 1,200, which is insane. And this is something that you can iterate on further. But in terms of its design capability, it's a lot better than the previous checkpoints that we've seen. And it's very similar in terms of what GBT5 is capable of generating. If we go back to the GBT 5 high generation, it is really similar in terms of its design skill, I guess. And in terms of structure though, it's not as bad as what we see with the GPT5 cuz with the generation over here, it is well structured and it's something that has functional components.
And I realized something that I'm mostly getting generations from the Firefly model. This is where I requested it to create a futuristic SAS landing page which looks absolutely amazing in terms of its UI components. You can see that it looks really good. It added basic functions of a SAS landing page like the pricing tier, testimonials as well as the FAQ which all work and it looks absolutely amazing.
If you like this video and would love to support the channel, you can consider donating to my channel through the super thanks option below. Or you can consider joining our private Discord where you can access multiple subscriptions to different AI tools for free on a monthly basis, plus daily AI news and exclusive content, plus a lot more. But that's basically it, guys, for today's video on these new checkpoints. Let me know what you guys think. In my opinion, this does seem to be uh maybe the new upgraded GPT5 variants and it's something that will be released to counter maybe the Gemini 3.0 Pro model. It is something that both of these two model providers tend to do. Gemini and OpenAI, they tend to release different models whenever their competitor is releasing something big. So, this could be a hedge over whatever Gemini is ending up releasing in December.
But that's basically it, guys. Hope you enjoyed today's video and got some sort of value out of it. I'll leave all these links in the description below, but make sure you go ahead and subscribe to the second channel if you haven't already. Join the newsletter, join our private Discord, follow me on Twitter, and lastly, make sure you guys subscribe, turn on notification bell, like this video, and please take a look at our previous videos so that you can stay up to date with whatever is happening in the world of AI. But with that thought, guys, have an amazing day, spread positivity, and I'll see you guys really shortly. Peace out.