GPT 5.6: Sun, Earth, Moon Models Explained
44sThe naming and tiered model system (Sun, Earth, Luna) is intriguing and sparks curiosity about which model is best for different tasks.
▶ Play ClipOpenAI has released GPT 5.6 with three models—Sun, Earth, and Luna—alongside a new desktop app and advanced voice features. This video demonstrates the voice model's real-time translation, the Codex integration for coding, and the ability to generate games and websites with a single prompt.
OpenAI launched GPT 5.6 with three models: Sun (strongest), Earth (balanced), and Luna (lightest/cheapest). New features include a desktop app and improved voice interaction.
Free users get Earth with limitations; paid plans access more advanced models. Sol Ultra is not yet available to everyone. The models use fewer tokens, reducing quota consumption.
The voice model can act as a real-time translator, speaking naturally with pauses and noises. It decides when to stop or speak, making conversations feel human-like.
The new desktop app integrates Codex, allowing users to ask for tasks, access the desktop, and install plugins for favorite applications. All models are selectable with adjustable effort levels.
Using a single prompt, the Sun model created a 3D HTML game (pelican collecting coins) with responsive controls. The game was generated autonomously while the intro was being made.
The Terra model created a website with space animation style, displaying information about the models. It used research data and a video script as context.
OpenAI's Atlas browser is retired; instead, users can install a Chrome extension for browsing. GPT 5.6 handles thinking, voice chat, and tasks, while Codex handles coding.
GPT 5.6 introduces powerful new models and features that significantly enhance productivity, from real-time voice translation to autonomous code and content generation. The video showcases impressive capabilities that promise to streamline workflows for developers and content creators.
"The title accurately reflects the release of GPT 5.6 with human-like voice, but exaggerates 'speaks like a human' slightly."
What are the three models in GPT 5.6?
Sun (strongest), Earth (balanced), and Luna (lightest/cheapest).
01:03
Which model is available to free users?
Earth, with limitations.
01:49
What is the name of the new desktop app integrated with GPT Chat?
Codex.
06:21
What did the Sun model create with a single prompt?
A 3D HTML game where a pelican collects coins.
07:57
What happened to OpenAI's Atlas browser?
It was retired; users now use a Chrome extension for browsing.
10:38
Real-Time Voice Translation
Demonstrates the voice model's ability to translate languages in real time with natural pauses and noises.
03:15Autonomous Game Creation
The Sun model generated a fully playable 3D game from a single prompt, showcasing advanced autonomous coding.
07:57Context-Aware Website Generation
The Terra model created a themed website using provided research and script context, highlighting the importance of context.
09:33[00:03] just been released, and along with it, several new features have been launched that you 'll find very interesting. If you use GPT chat, if you use OpenAI tools, this will really help you a lot in your day-to-
[00:18] day work. They released a new computer application and also a new talking to a human, in addition, of course, to the three models: Sun, Earth, and Moon or Luna. In this video, I want to show you
[00:32] a little bit of everything that has just been released, how you can some tests. So watch this video until the end, because I'm sure n't be left behind. I'm Rob. Here we talk about Canva,
[00:47] tools that can help you be much more productive and make money online. If subscribe to the channel here. We have videos like this every week, okay? Let's from OpenAI. They launched it yesterday and they launched it again
[01:03] today. The highly anticipated launch of this new model, which everyone who uses artificial intelligence has probably already heard of, is the GPT 5.6 chat. GPT 5.6, the Sun, Earth and Moon (or
[01:18] Luna) model. Each of these models does something different. The Sun is the strongest model of all, the Earth provides balance, and the Moon is the lightest and cheapest model. For those who use it for programming, here's the API pricing, which is what you
[01:32] pay per use. I have their plan, a plan that I pay $ per month for, so I can access these models here, but several companies use them through API keys that they get directly from OpenAI. And here are
[01:49] the prices, but we're not going to focus on that here because the idea is to quickly show you the access details. Everyone who has the free GPT will have access to the Terra model, obviously with the necessary limitations. And for
[02:03] paid plans, you'll already have access to the more advanced models. Obviously, there's a model called Sol Ultra. This model isn't available to everyone yet. Apparently, the models are much smarter and use far
[02:17] fewer tokens. In other words, if you're using the model to schedule and perform tasks, you won't be constantly hitting your quota. Compared to Fable 5, which is the Antropic model, and which was even going to be removed from
[02:33] paid plans on the 7th, and Antropic, for some reason, I believe it's until the 12th. It's a fantastic model, I 'm using it daily, but it's quite expensive. He is capable of programming for you on his own for days.
[02:47] So, I've already seen some tests here from people who already had access, who you can ask and it will program it for you. So, it's as if there are several agents, all with the same power, working at the same time. He can
[03:01] run up to four versions of himself on a difficult task and fire up several other agents. Obviously, the cost will be higher if the person does that. So, I want to show you here, starting with the voice model. If you
[03:15] update your app, you'll be able to access the new voice model. Remember that if you have the free version, you'll have access to the mini model, which isn't as responsive, but it's still better than the
[03:28] Hello everything is fine? All very well. How can I help you today? Let's go. Look, I'm showing my audience the new voice model, and I'd like to ask you to act as a
[03:43] real-time translator. So, I'll speak in English and you'll translate everything into Portuguese in real time. Could it be? Understood? I'll start the now. Ok. So,
[03:57] Everything I say in English, you will Everything I say in English, you will translate into Portuguese. with GPT and it's very good to use.
[04:13] I 'm eager to see what I can create with this. I want to know what they think. I want to know where they are from and ask them to leave
[04:28] a comment stating what city they are from. They can comment below. OK. You can now translate what I say in Spanish into Portuguese. Perfect, understood. I will be translating from
[04:42] Spanish to Portuguese in real time from now on . Now I'm going to speak Now I'm going to speak a little bit in Spanish. So I can. So I can make mistakes.
[04:57] I prefer speaking English. I prefer speaking English. Check this out, guys. I used it here as a translator, but you can keep the conversation going, you can interrupt, you can talk. They are, wow, it's
[05:12] be very useful, especially for those who are real-time translator here, as you just saw there. It's phenomenal. And this new model has a major advantage: it
[05:27] decides whether to stop, whether to speak, and it even makes the little noises you hear when really weird, man. It's about hearing and knowing that you're talking to your phone. Artificial intelligence is increasingly resembling that movie where
[05:40] artificial intelligence, isn't it? And you can ask difficult questions, and it you and bring you that information. Fantastic. But for those concerned
[05:52] about behavior, OpenI itself said that the model is trained to recognize certain speech patterns and certain questions, obviously to prevent future problems. So I believe they are always very
[06:05] concerned about safety. And then we'll move on to the part where we need to respond so we can execute the process. And I want to show you the new app here. So now when you download the GPT application to your computer, it will be the Codex.
[06:21] Then the Codex integrated with the GPT chat. So now you can simply come here and ask for a task. So, if you're going to use it on a daily basis, you use it here for work. This one is for everyday use. Here at the bottom,
[06:33] am with the Sol model, see? All models are available here. You select the effort level of the model. The higher the altitude, obviously, the more it will consume. I'm here with access to ultra. Hmm, I thought I didn't have
[06:45] access, but it's already available to me. Speed too. And here you else. But in addition to that, you can also do the following. Just look. applications, you can ask the computer to access your desktop,
[07:00] just like in Cloud Cowork, where you can ask it to access it can do that too. He's going to get here, access the computer, and do plugins available here for you to install. So, you can install
[07:14] plugins for your favorite applications and ask GPT to interact with those applications. This is only for the desktop version, okay? your scheduled tasks, your projects, and so on. So, if you're interested and you work with
[07:29] my job, not here on my personal computer; I have a separate account for that. Now you have the potential to keep everything together in a single application. The capacity is simply fantastic. We're not going to explore them all
[07:44] here, partly because we don't have time to do that, but I'll show you several ways to use them. I asked him here, in the tasks section, to create a 3D game for me while I was
[07:57] making the intro for this video, okay? I ordered the sun model here, and I set it to auto. I haven't seen that here yet. I'm going to open it. And I asked him to create this game, to make it an HTML game that I could
[08:10] open without needing to install anything. So, I'm going to click here. The game has begun, man. Look at this, man. [laughs] Yeah, yeah, yeah, amazing. Let me open it here in the window. Okay, it opened here. So
[08:24] pelican. This pelican here, for those who do n't know, those who follow artificial intelligence know that they always do tests with a pelican and such. So, look at this, man. Look at this. Does it make a little noise? I don't think there's any little
[08:36] noise, look. And he's picking up the coins, look. He hit it there. Man, this is amazing and very responsive. He created this entire game with just one prompt. I gave him a prompt, and he created this entire game. And here I have space
[08:51] to jump, look. I jump with space and start earning points there. I mean, obviously it's not a complete game, but man, knowing that this was created with just one prompt is simply fantastic. And then you can keep
[09:05] Just imagine what it's like for someone who works in development. Wow, man, I'm super excited to see what I can do with this tool, and comparison with Fable. which is the other tool. But wow, it's incredible, you know?
[09:20] Very cool. I didn't ask for anything to be added, sound effects and all that, but I can ask for it here. But you can see that it works very well compared to other models. That was the sun. The sun did this for me. So I asked the
[09:33] Terra model to create a website with the information I gathered, the research I did, so I could bring this video to you all here, and also my video script. So I gave him the freedom to create the website and asked him to create it in a
[09:46] style that looked like a space animation. I got this as inspiration from the Open Aa presentation itself, and he made this website, see, he brought all the information here. Here he shows what the models are and stuff,
[09:59] talks and I don't know what else. Here, you can select the Sun, Earth, and Moon models. Wow, that's fantastic! One problem is multiple minds, when the brain becomes a controlled object. This section briefly discusses the issue of the blockade
[10:12] that occurred. Look at this. Now you can imagine how you'll develop it, right? That artificial intelligence, the more context you provide, the better you'll be able to deliver content. Obviously, the context is valid, right? So here, you can see that it
[10:26] valid, right? So here, you can see that it wasn't a prompt, and he created this entire website for me. Wow, fantastic. Now, the Atlas browser that OpenAI tried to create its own browser is no longer around; it didn't work out
[10:38] very well, and it's been retired. You can access your browser here through the OpenAI extension you install from Chat GPT, and it will navigate for you, just like the Antropic extension with Cloud. So now the
[10:54] Antropic extension with Cloud. So now the GPT 5.6 chat does the thinking, the live voice chat talks GPT 5.6 chat does the thinking, the live voice chat talks to you, and the GPT Work chat does things just showed you here. In addition, the codex is there so you can encode it. I do
[11:08] n't know if most of the people watching me here program or use this part of GPT. I use it a lot in my day-to- day work, I've been using the Cloud a lot, but this launch I'm super excited to see what I'll be able to
[11:23] build. So, in summary, if you have the free version of GPT, you'll most likely be using it between Terra and Luna, and you'll encounter some limitations. Comment below what you think. I have many
[11:39] content ideas I'd like to share with you on this topic, and I wanted to make this video to impressed. What is your opinion? Leave a comment below so I can
[11:53] also know which city you 're talking about, and if you're watching me, so I can go there and send you a hug. I hope you enjoyed the video. See you next time. See you later. Ciao. Ciao.
⚡ Saved you 0h 12m reading this? Transcribe any YouTube video for free — no signup needed.