×
Asus ROG Xbox Ally X Review: Everything’s an Xbox, Even When It’s Not

Asus ROG Xbox Ally X Review: Everything’s an Xbox, Even When It’s Not

Everything is an Xbox now, right? If that’s the case, why can’t we mod our smart fridge or smart toaster with screens to transform them into the rough approximation of an Xbox gaming machine? The hard truth is there is no unified sense of “Xbox” anymore. That’s epitomized by the $1,000 Asus ROG Xbox Ally X handheld. It’s the first piece of real gaming hardware with an Xbox name to roll out in five years. Despite the moniker, it’s far more of a PC than it is any flavor of Xbox console. And thanks to Microsoft’s efforts to fix Windows for a small screen, it’s one of the best handheld PCs yet.

The ROG Xbox Ally X can certainly compete with Valve’s Steam Deck, and it can do so without worrying about kernel anti-cheat compatibility. The one point it can’t compete on is price. At $1,000, the ROG Xbox Ally X will seem like an overindulgent splurge that won’t net you the ease or comforts of a true-blue console. And though it has better graphics potential than Valve’s handheld, maximizing its capabilities requires coaxing.

If this is what we can expect from “Xbox” going forward, then console gamers had better get used to their new PC playground—whether that means adjusting power and graphics settings or dealing with Windows’ penchant for bugs and missing settings. But it also means a cleaner, more gamer-focused experience than any other handheld Windows device released over the past several years can match.

Asus ROG Xbox Ally X

It’s the first step in a long road to reinventing the Xbox brand, but in the meantime its one of the best handheld PCs you can get, despite glaring power standby bugs.

  • Controller-like grips are comfortable
  • Performance at lower power
  • Windows finally good on handhelds
  • Access to all your game launchers
  • Better than average battery life
  • So-so screen and speakers
  • Missing Windows settings in FSE
  • Battery standby bug
  • More expensive than before

The real innovation of the Xbox Ally is the new “full screen experience,” or FSE. The built-in software finally alleviates the excruciating experience of forcing Windows 11 desktop to run on a 7-inch screen. Now, a hit of that newfangled Xbox button pulls up most (but not all) of the features and apps you need. Even better, you can access games across the most popular game launchers with a few quick taps. Plus, the device promises better performance on this version of Windows. In our tests, that proved true compared to devices with similar specs.

The new UI is imperfect, and the software is far from complete. You’ll still find game launchers or incongrous settings that require you to muddy your screen with your grubby fingerprints. There are options missing—some are still stuck behind Windows settings—and I’ve experienced concerning issues with the device depleting standby power, as if leaking juice like sand through an hourglass. Since this handheld has the entire weight of Microsoft’s gaming arm behind it, I hope updates come swiftly and that the FSE makes its way to more handhelds than just the one bearing Microsoft’s Xbox logo.

Still, this isn’t the Nintendo Switch 2 of the Xbox brand. The Xbox Ally is the kind of device to make me imagine what a true merging of PC and Xbox could be. It could prove expensive disaster—or a glorious (but still expensive) shift to the entire gaming landscape.

It’s kind of an Xbox, but it’s mostly an Ally

The most obvious difference between the old and new Ally X is the controller-like grips. © Adriano Contreras / Gizmodo

Xbox told me that Microsoft and Asus started working together shortly after the launch of the Asus ROG Xbox Ally X to craft the new handheld. What, can’t you tell? It’s the two prongs on both sides of the device that the two partners in crime claim was modeled directly on the Xbox wireless controller. When you hold the controller and Xbox Ally up together, you’ll see the similarity isn’t 1-to-1—not that it matters. Each prong holds a rumble motor that feels far more present than those of other unibody handhelds do.

Despite all that, I wouldn’t go as far as to say it’s more ergonomic than flat body designs. I can curl my digits around a Steam Deck and play comfortably just as well as I can with the ROG Xbox Ally X.

This is still a PC. With that in mind, you have access to two USB-C ports (one being USB-4 compatible, for hooking up external devices like an eGPU) on top alongside a microSD slot and a headphone jack. There’s no port on the bottom to enable a Switch-like dock. In fact, all your regular Steam Deck docks you bought for other handhelds may not work as well with the two grips sticking out from the body. My Steam Deck Dock from Gulikit does not fit the Xbox Ally even though it’s perfectly usable with past handhelds.

While I didn’t find the grips supremely comfortable, there’s another extra benefit to the controller-like design. My wrists never fell asleep as I lost circulation while playing while lying down like they often do when I’m using my personal Steam Deck of Switch 2. I like to play my handhelds in bed, so the extra grips proved to be a plus despite making the handheld even thicker.

It weighs in at 1.58 pounds, which is barely any more heft than the original $800 ROG Ally X and lighter than other handheld PCs like the Legion Go lineup.

The triggers and bumper buttons also have a feel akin to the Xbox controller and its “Impulse” triggers with individual vibration motors. The sides of the device are sloped more than they are on the ROG Ally X, meaning I don’t need to perform any crab claw gestures to access the LB and RB buttons. But if you were expecting the device to feel exactly like an Xbox controller, you’ll end up disappointed. The A,B,X,Y layout is the same, but the look and feel of both the buttons and joysticks are far more ROG Ally than Xbox.

In the end, all that matters is if it feels good to play. I can comfortably say: It does.

FSE is a great idea but still a work in progress

Asus Rog Xbox Ally X 17
Microsoft’s Windows “full screen experience” finally lets you navigate through different games and apps without needing to tap on a desktop. © Adriano Contreras / Gizmodo

Beyond the twin menu buttons, the only switch you need to care about for menu navigation is the new Xbox button found next to Asus’ own Armoury Crate toggle. Hitting the button once brings up the Game Bar menu, delivering quick access to your game library across all launchers. Hitting the library button will take you to the Xbox app where you can access Game Pass (you get a three-month Xbox Game Pass Premium subscription with the device; just don’t forget to cancel) and Xbox Cloud gaming. It’s here you’ll also find your library of both games and supported apps, which mostly include game launchers.

Xbox gamers who have never had to even consider their consoles’ power draw will now need to start thinking about TDP, or thermal design power. Despite originally telling me it was working on a replacement for power management. Microsoft left Asus’ Armoury Crate software in this handheld. You can access the program from the Game Bar to tweak the tpower draw you’re sending to the processor. More TDP means better performance at the cost of battery life. As a rule of thumb, you can stick to lower performance modes for two-dimensional or less intensive games. For everything else, shift it to “Turbo,” which ends up at 25W on battery power, or the max 35W if you have the device plugged in.

Asus Rog Xbox Ally X 13
Holding down the Xbox button grants you easy access to all your open windows. © Adriano Contreras / Gizmodo

How does it fare compared to Valve’s SteamOS, the Linux-based operating system made popular by the Steam Deck? Like SteamOS, the Windows FSE is a controller-first layout designed to make accessing the Xbox game store and your game library easier than it is on a desktop. Players like Asus, Lenovo, and MSI designed front ends to enable some easy access to your titles, but the FSE has come closest to making it feel more “console-like.”

When you hold down on the Xbox button, you can easily swap between all your open apps. Want to access Steam Big Picture Mode? It’s a few taps away, with no need to muddy the touchscreen with your fingerprints. It’s here you’ll also see an option to open up desktop mode if there are apps you can’t access through the Xbox app. Asus originally described the FSE to me as the system dual booting Windows with the FSE and regular desktop mode. However, the improved performance is available in either variation.

As if you needed another reminder, this is a PC, not an Xbox console. A 2023 update redesigned the Xbox app so it could be navigated with a controller. There’s still more work to be done. You can’t use the right stick to scroll. There isn’t enough use of the RB and LB buttons for navigating through side-by-side menus. Beyond those hangups, what’s particularly egregious is just how limited Microsoft made its first-party software. On Linux-based SteamOS, I can drop into the desktop mode, install Emulation Station, and access it from the Steam Deck’s home menu. The Xbox app will recognize apps like GOG Galaxy, Epic Games Store, Battle.net, and Ubisoft Connect, but it won’t enable other apps you may install on your system. Accessing them requires entering the Desktop Mode. SteamOS lets you modify it enough to enable third-party apps to appear on the homescreen, but Microsoft, being Microsoft, won’t let you do the same with the FSE.

Don’t be afraid to lower the TDP

Asus Rog Xbox Ally X 09
Games like Hollow Knight: Silksong run buttery smooth on the ROG Xbox Ally X, but it’s let down a bit by the display. © Adriano Contreras / Gizmodo

AMD’s Ryzen Z2 Extreme offers only a marginal bump in performance compared to the Ryzen Z1 Extreme that hit the scene two years ago. However, the updates Microsoft made to Windows 11 supposedly limit background tasks and boosts performance. From what I experienced with the chip on the Lenovo Legion Go 2 compared to the ROG Xbox Ally X, that seems to check out. This device has less RAM and still supported small performance gains at the same resolution and TDP, though it depends on the game how much better it can be.

The Lenovo Legion Go 2 set the new record for AMD Ryzen Z-series devices with its astronomical $1,350 launch price with the AMD Ryzen Z2 Extreme chip. It’s the same APU—or accelerated processing unit—found in the ROG Xbox Ally X that costs $350 less. The Legion Go 2 contains 32GB of RAM and sports a massive 8.8-inch OLED display. After spending more than a week with the Xbox Ally’s 7-inch IPS LCD, I miss those deeper black levels and contrast. For the sake of such an upgrade, you have to drop the price equivalent of a new Nintendo Switch—$350 after recent price hikes—for the larger handheld.

The Xbox Ally X comes with 24 GB of RAM and 1 TB of SSD storage. Along with its 80Wh battery, it’s equivalent in specs to last year’s ROG Ally X, but it now sports the new processor. I played the device for multiple hours at a time and never felt the device get hot under my hands. The fans can get loud, but not to the egregious jet airplane levels of larger devices like the Legion Go 2. The Z2 Extreme is an 8-core processor promoting better graphics performance, but you may not see such a big increase if you’re used to devices from the past two years.

At 1080p resolution and running at the max 35W when plugged in, I could get around 52 FPS in Cyberpunk 2077 running at Steam Deck settings. In actual gameplay, you’ll see framerates at just under 50 fps in most scenarios. At 30W, last year’s ROG Ally hits around 46 fps. Other games, like Shadow of the Tomb Raider, see better uplift, with the new Xbox Ally X netting around 10 fps more than its predecessor.

Synthetic benchmarks may help us get a better idea of how the device compares. I found the Xbox Ally X netted around 400 points more in 3DMark Steel Nomad Light and 600 points better in 3DMark Time Spy. In actual gameplay, the difference is more minute, especially at the highest wattage. In my tests with games like Baldur’s Gate III, the device didn’t perform exceptionally better than the MSI Claw 8 AI+ running on an Intel Core Ultra 258V chip from last year in 3DMark tests. In others, the Ryzen Z2 Extreme underperformed Intel’s laptop chip. The rat race for better performance ignores how small improvements mean a lot more on a power-sipping device like a handheld, where 5 fps is the difference between a game being playable or not.

The Z2 Extreme’s hidden secret is that it performs better at lower wattages than previous generations. Tests at lower wattages in multiple games, relying often on AMD’s FSR upscaling, showed I could finally achieve playable framerates in games like Indiana Jones and the Great Circle and Star Wars Outlaws. Even on “Turbo” power mode, the device will default to 25W when not plugged in. In that way, I found I could play multiple current AAA games (and a few unreleased Xbox titles) at playable frame rates with some finagling of graphics settings.

Xbox claimed the big update alongside the full-screen experience greatly enhanced performance, and that more than proved true. The secret is, you don’t even need the new UI to experience the performance boost. I updated the Lenovo Legion Go 2 to the latest Windows Insider Update without bothering to install the updated user interface. The uptick in frame rate in some games could move the needle by 5 fps or more in games like Cyberpunk 2077, though the increase could be as much as 3 fps in titles like Shadow of the Tomb Raider. In benchmarks, the upgrade was so dramatic that Lenovo’s latest handheld eclipsed the ROG Ally X by 50 to 100 points in some tests. The Legion Go 2 holds more RAM than the new ROG handheld, which is enough to equal the slightest edge in overall performance.

There’s a voice in the back of my brain that’s hard to ignore. It tells me to shake my fist at Microsoft in a “man yells at clouds” fashion over how simple this fix seems to have been. Are you telling me we could have had better game performance on Windows this whole time?

The screen is just good enough

Asus ROG Xbox Ally X handheld hands-on
Yes, I miss OLED, but not enough to pay an extra $350 for a Lenovo Legion Go 2. © Raymond Wong / Gizmodo

There’s a “just fine” quality to the ROG Xbox Ally’s 1080p display. It’s a semi-glossy IPS LCD that doesn’t have the same “inky” qualities of OLED. And still, it’s serviceable in the same sense that Apple’s MacBook displays get the job done, even if it’s the same damn screen every damn year. I’m tired of the large bezels surrounding the display, and the size makes it difficult to read text in games like Warhammer 40K: Rogue Trader and other RPGs. Going from the Legion Go 2 to the Xbox Ally X in quick succession helped me realize how a bigger screen is sometimes worth the extra heft.

Just like the 2023 ROG Ally and 2024 ROG Ally X, the screen maxes out at 120Hz refresh rate. That’s more than enough for 3D titles. Then, when you opt to play a game like Hollow Knight: Silksong or Hades II, and your game can easily net over 160 fps on the lowest power settings, you may start to feel a little cheated at the minimum refresh rate without VRR (variable refresh rate) after paying $1,000 for a handheld like this. The screen is plenty bright indoors and in ambient light, but it’s still not enough to survive direct sunlight, either.

Sound quality through the dual speakers wasn’t something to write home about either. The sound can get plenty loud, but unlike the Legion Go 2 I felt I wanted to reach for a pair of gamer headphones or earbuds more often. There’s nothing terrible about the screen or audio quality. It just feels lackluster considering this is the third iteration of Asus’ handheld design, now for an inflated price.

The Xbox Ally X has concerning battery standby issues

Asus Rog Xbox Ally X 18
Don’t expect to play the ROG Xbox Ally X for too long, depending on the game. © Adriano Contreras / Gizmodo

Handhelds don’t have great battery life. Even thought the Xbox Ally X with its 80Wh battery can claim better play times than other handhelds, it would be barely enough to survive a two-and-a-half-hour plane flight. Running the device at lower power playing a game like Hades II, I could get closer to three and a half hours of battery life before the device was begging for the sweet embrace of a power cable. With more intensive games, that battery life could be cut down to around two and a half hours.

It’s better than some handhelds. The Lenovo Legion Go 2 and MSI Claw 8 AI+ with their larger screens and higher resolution will sap energy faster. The Switch 2 has notoriously poor battery life, and you’ll be lucky to get two hours out of the device before it’s through. However, the Xbox Ally X has other problems when trying to put the device to sleep. There were times I put the device to sleep by hitting the power button, but then I would re-enter into a gray screen, requiring me to restart the game.

I also found multiple instances when I hit the power button to put the device to sleep, only to find it was still sucking power for hours, leaving the device dead once I picked it back up again. The fans would occasionally kick in while asleep, which indicated to me there’s an issue Microsoft needs to fix before I can safely store the device away after a quick gaming session. It’s clear Microsoft was working on this software until the last minute, and that means there’s still work needed to clear up these glitches. The issues were enough to take what would otherwise be a glowing review and diminish it.

What should we eXpect from Xbox?

Asus ROG Xbox Ally X handheld hands-on
I’ll never go back to the days of needing to swipe on the desktop. © Raymond Wong / Gizmodo

High-end handhelds now cost more than $1,000. Sorry, but that’s what we should expect from now until Sony releases a handheld PlayStation 6. The base ROG Xbox Ally in white has a lower-end Ryzen Z2A chip I have not yet had the chance to test in full. That device is made to compete with the $550 Steam Deck OLED or Lenovo Legion Go S with SteamOS. Inevitably because of tariffs and trade retaliations from China, we can’t expect any handhelds to truly be cheap unless you aim for Nintendo’s more restrictive hybrid handheld console.

There’s a part of me that still suggests handheld buyers should hold off for the next generation of handhelds. Today’s Z2 Extreme devices are running on a GPU with AMD’s RDNA 3.5 microarchitecture. The next chip may use RDNA 4, which would give devices access to AMD’s FidelityFX Super Resolution 4. That would offer improved AI upscaling for games, improving performance by running the game at lower resolutions while making them appear as if they’re running at a higher resolution. It’s what future consoles may rely on for better optimized games. Microsoft and Sony will likely push developers to use the technology for its consoles. The Xbox Ally X will be left out of the loop.

Either way, Microsoft seems keen to work on the FSE to improve the experience in preparation for a next-gen Xbox console. Hopefully, that means Xbox has enough incentive to fix what’s broken and polish the experience to a squeaky-clean shine. Is it enough to make me abandon the Steam Deck? Maybe one day. Is it enough to give Xbox gamers the new hardware they’re looking for? It’ll depend on their tolerance for the PC gaming lifestyle.

Source link
#Asus #ROG #Xbox #Ally #Review #Everythings #Xbox

These days, new versions of AI chatbots don’t just launch; they’re unshackled and released to the public following government scrutiny. OpenAI’s new GPT-5.6 models were – like Anthropic’s Claude Mythos and Fable – apparently too powerful to just launch; but now, after some tinkering, they’re available to you, dear customer.

In practice, it simply means that the new GPT-5.6 models are very powerful and smarter than before. In its introductory post, OpenAI shared a bunch of graphs showing just how much better GPT-5.6 is than the competition, whilst using fewer tokens and generally costing less.

OK, great. But GPT-5.6 is not just one model; it comes in three distinct flavors: Sol, Terra, and Luna. So what do different kinds of users get, what should they pay for, and which models should they (mostly) use? Let’s dive in.

Free users get (almost) nothing

Sorry; if you’re not a paying customer, you’ll have to make do with OpenAI’s previous flagship model, GPT-5.5. Any sort of access to GPT-5.6 models requires a subscription of some sort. Fortunately, GPT-5.5 is still quite capable at most tasks, but if you want the best of the best, you’ll have to cough up the dough.

There’s an exception to this: Free and Go users can access GPT-5.6 through ChatGPT Work. More on that below.

If you’re a Plus or Business user, you can only get Sol (the most powerful model) at medium and higher effort settings. There’s another, higher level of performance called Sol Pro, but that’s only available for Pro and Enterprise users.

In terms of availability per one million tokens, the prices are: $5 input and $30 output for Sol. $2.5 input and $15 output for Terra, and $1 input and $6 output for Luna.

Sol, Terra, or Luna?

Why are there three models in the first place? Well, OpenAI always had a multi-tier model; for example, previously users were able to choose a “mini” version of the main model to get results done cheaper. Now, the model has been split into three tiers.

If you’re a paying customer, you’re free to use all three. But you know how it is in the world of LLMs: If you pick the smartest one, your usage limits will get hit faster (yes, there are always usage limits, even if you throw a ton of money at OpenAI).

In the simplest sense, GPT-5.6 Sol is the smartest model, Terra is in between (with roughly GPT-5.5 level of performance), and Luna is the cheapest, fastest, but also least capable of the bunch.

The breakdown is as follows: Terra is a “balanced” model for everyday work. That’s the one you should be asking most of your questions. Don’t underestimate it, though, as OpenAI claims it performs better than Anthropic’s Fable 5 in some cases.

Luna is cost-efficient, and should be used for easy, non crucial everyday tasks; think recipes and movie recommendations. Again, OpenAI says it outperforms Anthropic’s Opus 4.8 in some cases, so it’s not a slouch, either.

Sol should be reserved for coding, deep research, planning, and cybersecurity: The most demanding tasks. Of course it comes at a (literal) cost: While OpenAI claims it spends less tokens than Anthropic’s Fable 5, Sol will still hit usage limits a lot faster than the other variants.

Fun fact: If you ask GPT-5.5 about any of this, right now, it’ll give you completely wrong answers. Hopefully OpenAI will fix this soon.

Wait, what’s this ChatGPT Work thing, then?

Oh yeah, OpenAI also launched ChatGPT Work, which is a new agent in ChatGPT that can access and take actions on your apps and files, and work in the background until a task is finished. It’s powered by Codex (OpenAI’s software engineering agent) and GPT-5.6. Think about it as your buddy that will go through your emails and files, browse the web, fetch the relevant data, and create that presentation your boss wants before the day is done.

ChatGPT Work is rolling out to Pro, Enterprise and Edu users first on web and mobile; this will be expanded to Plus and Business users “over the next few days.”

On the desktop, Work is available for everyone, including Free users.

Oh, and one more thing: The fact that ChatGPT Work has a built-in browser also means that OpenAI is sunsetting its standalone web browser, Atlas. Sorry.

How about GPT Live?

GPT Live is a new version of ChatGPT Voice and it will show up when you start talking to ChatGPT.

We’ve covered this in more depth here, but the bottom line is that GPT Live can listen and speak at the same time, allowing it to keep up a more realistic conversation.

Wrapping it all up

The new GPT-5.6 model is smart. It comes in three flavors: Luna, Terra, and Sol, with Sol being the most capable variant, Luna the most affordable one, and Terra somewhere in the middle. You can currently only get them on paid tiers, unless you’re using ChatGPT Work on desktop. And ChatGPT Voice has also gotten smarter with GPT-Live underneath, a model that can listen and speak at the same time.

#GPT5.6 #Sol #Terra #Luna">GPT-5.6 Sol, Terra, and Luna are here. See which one’s best for you.
                                                            These days, new versions of AI chatbots don’t just launch; they’re unshackled and released to the public following government scrutiny. OpenAI’s new GPT-5.6 models were – like Anthropic’s Claude Mythos and Fable – apparently too powerful to just launch; but now, after some tinkering, they’re available to you, dear customer.In practice, it simply means that the new GPT-5.6 models are very powerful and smarter than before. In its introductory post, OpenAI shared a bunch of graphs showing just how much better GPT-5.6 is than the competition, whilst using fewer tokens and generally costing less. 


OK, great. But GPT-5.6 is not just one model; it comes in three distinct flavors: Sol, Terra, and Luna. So what do different kinds of users get, what should they pay for, and which models should they (mostly) use? Let’s dive in. 
Free users get (almost) nothingSorry; if you’re not a paying customer, you’ll have to make do with OpenAI’s previous flagship model, GPT-5.5. Any sort of access to GPT-5.6 models requires a subscription of some sort. Fortunately, GPT-5.5 is still quite capable at most tasks, but if you want the best of the best, you’ll have to cough up the dough. There’s an exception to this: Free and Go users can access GPT-5.6 through ChatGPT Work. More on that below. If you’re a Plus or Business user, you can only get Sol (the most powerful model) at medium and higher effort settings. There’s another, higher level of performance called Sol Pro, but that’s only available for Pro and Enterprise users. In terms of availability per one million tokens, the prices are:  input and  output for Sol. .5 input and  output for Terra, and  input and  output for Luna. 
        SEE ALSO:
        
            Visa is connecting with ChatGPT to let AI agents automatically make purchases
            
        
    
Sol, Terra, or Luna?Why are there three models in the first place? Well, OpenAI always had a multi-tier model; for example, previously users were able to choose a “mini” version of the main model to get results done cheaper. Now, the model has been split into three tiers. If you’re a paying customer, you’re free to use all three. But you know how it is in the world of LLMs: If you pick the smartest one, your usage limits will get hit faster (yes, there are always usage limits, even if you throw a ton of money at OpenAI). 
        
            Mashable Light Speed
        
        
    
In the simplest sense, GPT-5.6 Sol is the smartest model, Terra is in between (with roughly GPT-5.5 level of performance), and Luna is the cheapest, fastest, but also least capable of the bunch. The breakdown is as follows: Terra is a “balanced” model for everyday work. That’s the one you should be asking most of your questions. Don’t underestimate it, though, as OpenAI claims it performs better than Anthropic’s Fable 5 in some cases. Luna is cost-efficient, and should be used for easy, non crucial everyday tasks; think recipes and movie recommendations. Again, OpenAI says it outperforms Anthropic’s Opus 4.8 in some cases, so it’s not a slouch, either.Sol should be reserved for coding, deep research, planning, and cybersecurity: The most demanding tasks. Of course it comes at a (literal) cost: While OpenAI claims it spends less tokens than Anthropic’s Fable 5, Sol will still hit usage limits a lot faster than the other variants. Fun fact: If you ask GPT-5.5 about any of this, right now, it’ll give you completely wrong answers. Hopefully OpenAI will fix this soon.Wait, what’s this ChatGPT Work thing, then?Oh yeah, OpenAI also launched ChatGPT Work, which is a new agent in ChatGPT that can access and take actions on your apps and files, and work in the background until a task is finished. It’s powered by Codex (OpenAI’s software engineering agent) and GPT-5.6. Think about it as your buddy that will go through your emails and files, browse the web, fetch the relevant data, and create that presentation your boss wants before the day is done. 



ChatGPT Work is rolling out to Pro, Enterprise and Edu users first on web and mobile; this will be expanded to Plus and Business users “over the next few days.”On the desktop, Work is available for everyone, including Free users. Oh, and one more thing: The fact that ChatGPT Work has a built-in browser also means that OpenAI is sunsetting its standalone web browser, Atlas. Sorry. How about GPT Live?GPT Live is a new version of ChatGPT Voice and it will show up when you start talking to ChatGPT. 


We’ve covered this in more depth here, but the bottom line is that GPT Live can listen and speak at the same time, allowing it to keep up a more realistic conversation.Wrapping it all upThe new GPT-5.6 model is smart. It comes in three flavors: Luna, Terra, and Sol, with Sol being the most capable variant, Luna the most affordable one, and Terra somewhere in the middle. You can currently only get them on paid tiers, unless you’re using ChatGPT Work on desktop. And ChatGPT Voice has also gotten smarter with GPT-Live underneath, a model that can listen and speak at the same time.

                    
                                            
                            
    
        Topics
                    Artificial Intelligence
                    OpenAI
            

                        
                                    #GPT5.6 #Sol #Terra #Luna

Fable – apparently too powerful to just launch; but now, after some tinkering, they’re available to you, dear customer.

In practice, it simply means that the new GPT-5.6 models are very powerful and smarter than before. In its introductory post, OpenAI shared a bunch of graphs showing just how much better GPT-5.6 is than the competition, whilst using fewer tokens and generally costing less.

OK, great. But GPT-5.6 is not just one model; it comes in three distinct flavors: Sol, Terra, and Luna. So what do different kinds of users get, what should they pay for, and which models should they (mostly) use? Let’s dive in.

Free users get (almost) nothing

Sorry; if you’re not a paying customer, you’ll have to make do with OpenAI’s previous flagship model, GPT-5.5. Any sort of access to GPT-5.6 models requires a subscription of some sort. Fortunately, GPT-5.5 is still quite capable at most tasks, but if you want the best of the best, you’ll have to cough up the dough.

There’s an exception to this: Free and Go users can access GPT-5.6 through ChatGPT Work. More on that below.

If you’re a Plus or Business user, you can only get Sol (the most powerful model) at medium and higher effort settings. There’s another, higher level of performance called Sol Pro, but that’s only available for Pro and Enterprise users.

In terms of availability per one million tokens, the prices are: $5 input and $30 output for Sol. $2.5 input and $15 output for Terra, and $1 input and $6 output for Luna.

Sol, Terra, or Luna?

Why are there three models in the first place? Well, OpenAI always had a multi-tier model; for example, previously users were able to choose a “mini” version of the main model to get results done cheaper. Now, the model has been split into three tiers.

If you’re a paying customer, you’re free to use all three. But you know how it is in the world of LLMs: If you pick the smartest one, your usage limits will get hit faster (yes, there are always usage limits, even if you throw a ton of money at OpenAI).

In the simplest sense, GPT-5.6 Sol is the smartest model, Terra is in between (with roughly GPT-5.5 level of performance), and Luna is the cheapest, fastest, but also least capable of the bunch.

The breakdown is as follows: Terra is a “balanced” model for everyday work. That’s the one you should be asking most of your questions. Don’t underestimate it, though, as OpenAI claims it performs better than Anthropic’s Fable 5 in some cases.

Luna is cost-efficient, and should be used for easy, non crucial everyday tasks; think recipes and movie recommendations. Again, OpenAI says it outperforms Anthropic’s Opus 4.8 in some cases, so it’s not a slouch, either.

Sol should be reserved for coding, deep research, planning, and cybersecurity: The most demanding tasks. Of course it comes at a (literal) cost: While OpenAI claims it spends less tokens than Anthropic’s Fable 5, Sol will still hit usage limits a lot faster than the other variants.

Fun fact: If you ask GPT-5.5 about any of this, right now, it’ll give you completely wrong answers. Hopefully OpenAI will fix this soon.

Wait, what’s this ChatGPT Work thing, then?

Oh yeah, OpenAI also launched ChatGPT Work, which is a new agent in ChatGPT that can access and take actions on your apps and files, and work in the background until a task is finished. It’s powered by Codex (OpenAI’s software engineering agent) and GPT-5.6. Think about it as your buddy that will go through your emails and files, browse the web, fetch the relevant data, and create that presentation your boss wants before the day is done.

ChatGPT Work is rolling out to Pro, Enterprise and Edu users first on web and mobile; this will be expanded to Plus and Business users “over the next few days.”

On the desktop, Work is available for everyone, including Free users.

Oh, and one more thing: The fact that ChatGPT Work has a built-in browser also means that OpenAI is sunsetting its standalone web browser, Atlas. Sorry.

How about GPT Live?

GPT Live is a new version of ChatGPT Voice and it will show up when you start talking to ChatGPT.

We’ve covered this in more depth here, but the bottom line is that GPT Live can listen and speak at the same time, allowing it to keep up a more realistic conversation.

Wrapping it all up

The new GPT-5.6 model is smart. It comes in three flavors: Luna, Terra, and Sol, with Sol being the most capable variant, Luna the most affordable one, and Terra somewhere in the middle. You can currently only get them on paid tiers, unless you’re using ChatGPT Work on desktop. And ChatGPT Voice has also gotten smarter with GPT-Live underneath, a model that can listen and speak at the same time.

#GPT5.6 #Sol #Terra #Luna">GPT-5.6 Sol, Terra, and Luna are here. See which one’s best for you.

These days, new versions of AI chatbots don’t just launch; they’re unshackled and released to the public following government scrutiny. OpenAI’s new GPT-5.6 models were – like Anthropic’s Claude Mythos and Fable – apparently too powerful to just launch; but now, after some tinkering, they’re available to you, dear customer.

In practice, it simply means that the new GPT-5.6 models are very powerful and smarter than before. In its introductory post, OpenAI shared a bunch of graphs showing just how much better GPT-5.6 is than the competition, whilst using fewer tokens and generally costing less.

OK, great. But GPT-5.6 is not just one model; it comes in three distinct flavors: Sol, Terra, and Luna. So what do different kinds of users get, what should they pay for, and which models should they (mostly) use? Let’s dive in.

Free users get (almost) nothing

Sorry; if you’re not a paying customer, you’ll have to make do with OpenAI’s previous flagship model, GPT-5.5. Any sort of access to GPT-5.6 models requires a subscription of some sort. Fortunately, GPT-5.5 is still quite capable at most tasks, but if you want the best of the best, you’ll have to cough up the dough.

There’s an exception to this: Free and Go users can access GPT-5.6 through ChatGPT Work. More on that below.

If you’re a Plus or Business user, you can only get Sol (the most powerful model) at medium and higher effort settings. There’s another, higher level of performance called Sol Pro, but that’s only available for Pro and Enterprise users.

In terms of availability per one million tokens, the prices are: $5 input and $30 output for Sol. $2.5 input and $15 output for Terra, and $1 input and $6 output for Luna.

Sol, Terra, or Luna?

Why are there three models in the first place? Well, OpenAI always had a multi-tier model; for example, previously users were able to choose a “mini” version of the main model to get results done cheaper. Now, the model has been split into three tiers.

If you’re a paying customer, you’re free to use all three. But you know how it is in the world of LLMs: If you pick the smartest one, your usage limits will get hit faster (yes, there are always usage limits, even if you throw a ton of money at OpenAI).

In the simplest sense, GPT-5.6 Sol is the smartest model, Terra is in between (with roughly GPT-5.5 level of performance), and Luna is the cheapest, fastest, but also least capable of the bunch.

The breakdown is as follows: Terra is a “balanced” model for everyday work. That’s the one you should be asking most of your questions. Don’t underestimate it, though, as OpenAI claims it performs better than Anthropic’s Fable 5 in some cases.

Luna is cost-efficient, and should be used for easy, non crucial everyday tasks; think recipes and movie recommendations. Again, OpenAI says it outperforms Anthropic’s Opus 4.8 in some cases, so it’s not a slouch, either.

Sol should be reserved for coding, deep research, planning, and cybersecurity: The most demanding tasks. Of course it comes at a (literal) cost: While OpenAI claims it spends less tokens than Anthropic’s Fable 5, Sol will still hit usage limits a lot faster than the other variants.

Fun fact: If you ask GPT-5.5 about any of this, right now, it’ll give you completely wrong answers. Hopefully OpenAI will fix this soon.

Wait, what’s this ChatGPT Work thing, then?

Oh yeah, OpenAI also launched ChatGPT Work, which is a new agent in ChatGPT that can access and take actions on your apps and files, and work in the background until a task is finished. It’s powered by Codex (OpenAI’s software engineering agent) and GPT-5.6. Think about it as your buddy that will go through your emails and files, browse the web, fetch the relevant data, and create that presentation your boss wants before the day is done.

ChatGPT Work is rolling out to Pro, Enterprise and Edu users first on web and mobile; this will be expanded to Plus and Business users “over the next few days.”

On the desktop, Work is available for everyone, including Free users.

Oh, and one more thing: The fact that ChatGPT Work has a built-in browser also means that OpenAI is sunsetting its standalone web browser, Atlas. Sorry.

How about GPT Live?

GPT Live is a new version of ChatGPT Voice and it will show up when you start talking to ChatGPT.

We’ve covered this in more depth here, but the bottom line is that GPT Live can listen and speak at the same time, allowing it to keep up a more realistic conversation.

Wrapping it all up

The new GPT-5.6 model is smart. It comes in three flavors: Luna, Terra, and Sol, with Sol being the most capable variant, Luna the most affordable one, and Terra somewhere in the middle. You can currently only get them on paid tiers, unless you’re using ChatGPT Work on desktop. And ChatGPT Voice has also gotten smarter with GPT-Live underneath, a model that can listen and speak at the same time.

#GPT5.6 #Sol #Terra #Luna

Post Comment