The release went pretty much exactly as expected and has put us in profit for this month. We've enjoyed seeing people play on twitch and youtube and hearing all the feedback from you guys. A lot of it has already started to be implemented.
The backend went down once or twice. One time we accidentally DDOS'ed ourselves, and one time we had to scale some services to meet demand. As it usually is in these scenarios, the services that got overwhelmed weren't the ones we had considered at all.
Our current mixed review status (44%) wasn't unexpected. It's the reason we didn't want a frontpage takeover or any other push from Valve, because we wanted people to find this organically - not have it sold to them. The main complaints are AI Slop, Performance and This Isn't Garry's Mod. Not forgetting of course, our biggest fans got the game for free - so can't leave a review.
I want to talk through some of these complaints below.
The huge majority of games are created by smart, dedicated humans. So when the games are dismissed as "AI slop" it's really an injustice to the work those developers have put in.
Some games have their thumbnails created by AI and so get dismissed too. I don't think these guys are doing themselves any favours, but it's easy, so people do it.
AI is the teacher now. It's how people are going to learn how to program from now on. They're going to be generating their own examples and have it explain the code to them. I feel that outright banning the use of AI is a bad thing, because I'm not worried, I think human creativity is always going to win out.
We're not an AI based product in any way, but we live in the real world, so we're not anti-AI either. We're not going to put a pointless AI Assistant button for no good reason. We're not going to shove any AI tools down your throat. We're not gonna have a "make game" wizard that generates code for you. But if the best game in the world is made using the help of AI we're not going to tell you that you can't post it.
Our discovery algorithm isn't perfect right now but we're improving it every day. The main aim is that the good games float to the top and the bad games float to the bottom, while offering new, fresh content to people all the time.
One of the biggest things we're getting nailed on is performance, so I wanted to share where we are with it, what we're doing now and what our long term plans are.
If you're coming from Garry's Mod, things are going to feel different. Garry's Mod runs on a 20 year old engine that'll hit 300fps on a toaster - we're not there yet, and it's a fair comparison to make. We're aware that mid-range hardware in particular is having a rougher time than it should.
This is the worst the performance will ever be, we've been improving and will continue to improve. We know what the bad stuff is and we know how to solve it. We have a roadmap full of things we're actively working on to do it.
We sample performance data from the players across all varieties of hardware, and we display it all right here on sbox.game, this means when we fuck up you can hold us accountable, and when we improve we can gloat.Here you can see we've improved from a 17ms average to a 12ms average in the past 6 months, this is from looking at profilers, seeing the worst thing and improving it. Rinse and repeat.
Here's our average frame times we're seeing on the most popular low -> high hardware:
Ryzen 5 3600 + RTX 3060 = 54fps
Ryzen 7 7800 + RTX 4070 = 108fps
Ryzen 7 9800 + RTX 5080 = 125fps
There's a lot to improve on here across all hardware tiers, but it's not completely tragic. Each combination will have it's own set of limitations we'll have to solve.
Optimizing for a UGC platform carries its own unique challenges. We want games to be as simple to create as possible, without each creator having to over-worry about the low level performance of their game. We have to come up with solutions that cater to dozens of use cases from small to huge games of any genre.
Most modern engines spread rendering work across multiple frames. Temporal upscaling, temporal AA, amortized GI - it's fast but it all adds up to a blurry, ghosty mess that falls apart the moment you move.
We're not doing that. We render one frame every frame. It should look great in motion, not just screenshots. We're doing forward rendering, MSAA and everything in one frame.
This is a challenge, there are problems that are harder to solve, but we'd rather face those challenges than take the shortcuts other modern engines do.
A release to a wider audience always brings focus onto the most important things that we need to do.
We have development challenges ahead of us but this is what we're here for. This is what we love. We love it when things break because then we get to fix them. We want the opportunity to show everyone what we can do, we want that fight 💪🏼.
Congratulations once again to everyone at Facepunch and to everyone who’s been waiting for this moment on the release. Yesterday was an awesome day. I like everything about it, despite the issues and I’m confident that the next 10 years (just like the ones I spent on Garry’s Mod) will be incredibly strong for game development, since s&box is ushering in a new era: a platform and engine all in one (Epic is preparing to buy Unity and turn it into a platform for Fortnite as well, it's hard proof my words)
Yeah, back in 2023, I didn’t believe FP could pull off developing an engine and a platform either, and I thought: why are you doing this? gmod 2 is almost ready just release it. And as it turned out, they did it. And I’m sure that in three years, when S&Box becomes free, it’ll be even better, and in 10 years, when there are tons of new possibilities. So, thank you to Garry and the whole team for giving us the chance to realize our creative ambitions, bring joy to players, and make money for that.
Regarding the thoughts on AI Slop, I’m really glad that Facepunch is responding to this so well, without bias, and with the understanding that AI is just a tool that can produce both great and bad results. It can solve the design problem for programmers and the programming problem for designers.
🔥 I’m looking forward to gamejam date, and I’m making a game together with the whole community! Good luck to everyone.
It seems to me that over time, as higher-quality projects emerge, lazy AI slop will sink to the bottom of algorithms because most people simply won’t find it interesting. People will also just stop clicking on cringe AI thumbnails. The most important thing is that you keep doing your work and continue listening to the community. 20 year anniversary, lets fucking gooo!
smart decision on the "AI slop" discourse. Good games are good games. The players will decide what is good regardless of how it was created. Great work on the launch despite any hiccups!
Honestly huge respect for this post. The transparency alone is refreshing, admitting where things broke, explaining the reasoning behind decisions, and already shipping fixes this fast says a lot about the team.
Calling everything “AI slop” ignores how much real human work is going into these games, and I think your approach of not forcing AI but not demonizing it either is incredibly sensible. The performance breakdown was reassuring too. What stood out to me wasn’t just “we’ll optimize it,” but that there’s a clear technical philosophy behind what you’re building instead of taking easy shortcuts.
Mixed reviews aside, this made me more optimistic, not less. It feels like a passionate team building something for the long haul and actually listening. Excited to see where you take S&box. Once again, huge congrats to the team and thank you! <3
On my 5080, the game only gives about 120 unstable frames in 1440p and puts the GPU under full load. It would be great if you added DLSS Super Resolution and a frame generator to smooth things out.
What still pleasantly surprises me is that companies like Valve and Facepunch still exist today. In a way, they are two underground teams, even though the scale of their products makes that hard to say. They don’t feel like typical corporate machines; there is still a spirit of independent development, openness, and experimentation in them.
In that sense, Valve and Facepunch synthesize extremely well: on one side, there is infrastructure, Source 2, the Steam marketplace, and a massive audience; on the other, there is the experience of building living sandbox platforms where players become creators.
Of course, it’s not worth romanticizing them too much either. From the outside, leaders and companies can seem open, honest, and “one of us,” but we only see the public image. But it's working -_-
s&box has every chance to become a great success — or a great lesson.
You’re neglecting the sandbox game mode — that’s your core game. A new player will enter it first and explore the game through it. You need to develop it properly. Players are complaining that even the weapons don’t have reload sounds, and this game mode should be clearly visible and fixed in a prominent place in game hub.
Hello S&box team, I have been playing garrys mod since i was little. So far I really like this game because I never got to enjoy garrys mod when it was really famous. I always had to play on my cousins computer cause I never had anything of my own. This new s&box game means a lot to me and I love it. There are some things that I feel like the game can improve but I know they will be fixed. I have dreams of making my 1st ever game on here, and hope I can. I do hope you guys improve sandbox a lot and Hope it does not get hidden in the shadows.
I've been very impressed with performance networked or not-networked. To be able to zoom through massive cityscapes (with interiors) and with hundreds of NPCs and still get nearly 100 FPS possible (while in editor) on a 3060 was unexpected. This is largely thanks to GPU instancing, but also many other things. I can have multiple editors open, multiple VSCode IDEs, be running several servers, and not notice any major performance issues... that's impressive. Hot reloads work so well with the editor. There are so many things to be grateful for. Congratulations on the massive success.
The problem with the AI stuff isn't really that the game is being coded with AI stuff. I don't think people have a problem with that. It's 100% the AI thumbnails and generative AI use (as in, chat GPT style generative AI that actively steals from people, not procedural stuff that's been around for a bit).
I feel like you guys know this and are just hand waving away the issue for whatever reason. Tons of creative platforms are moving towards having generative AI labeled as such and allowing for filtering. I honestly don't understand exactly why you guys are seemingly so set in your ways with this one when it's been a very obvious complaint and issue for so long.
Just add a disclosure option for developers that says "this game contains AI generated content" and let us filter out those AI generated games if we don't want to support generative AI stuff. I really don't feel like this is a bad compromise at all, and it really is the most middle ground, easy W stance you could take and implement. Will it require effort? Yes, but it will produce a much better image for the game and people will be more willing to participate in the platform and support it. If your "main aim is that the good games float to the top and the bad games float to the bottom" then this should only help people find the games they're interested in and want to support.
I want to see this platform succeed. It's gotta feel human to succeed imo.
yknow I honestly wish there was a way to set any lobby to private, because I really don't want random person in stuff like for example the elevator/lift or something when I'm dicking around with friends.
Yeah, back in 2023, I didn’t believe FP could pull off developing an engine and a platform either, and I thought: why are you doing this? gmod 2 is almost ready just release it. And as it turned out, they did it. And I’m sure that in three years, when S&Box becomes free, it’ll be even better, and in 10 years, when there are tons of new possibilities. So, thank you to Garry and the whole team for giving us the chance to realize our creative ambitions, bring joy to players, and make money for that.
Regarding the thoughts on AI Slop, I’m really glad that Facepunch is responding to this so well, without bias, and with the understanding that AI is just a tool that can produce both great and bad results. It can solve the design problem for programmers and the programming problem for designers.
WW
The most important thing is that you keep doing your work and continue listening to the community. 20 year anniversary, lets fucking gooo!
Calling everything “AI slop” ignores how much real human work is going into these games, and I think your approach of not forcing AI but not demonizing it either is incredibly sensible. The performance breakdown was reassuring too. What stood out to me wasn’t just “we’ll optimize it,” but that there’s a clear technical philosophy behind what you’re building instead of taking easy shortcuts.
Mixed reviews aside, this made me more optimistic, not less. It feels like a passionate team building something for the long haul and actually listening. Excited to see where you take S&box. Once again, huge congrats to the team and thank you! <3
It would be great if you added DLSS Super Resolution and a frame generator to smooth things out.
In that sense, Valve and Facepunch synthesize extremely well: on one side, there is infrastructure, Source 2, the Steam marketplace, and a massive audience; on the other, there is the experience of building living sandbox platforms where players become creators.
Of course, it’s not worth romanticizing them too much either. From the outside, leaders and companies can seem open, honest, and “one of us,” but we only see the public image. But it's working -_-
s&box has every chance to become a great success — or a great lesson.
Strong, Garry. Strong.
I feel like you guys know this and are just hand waving away the issue for whatever reason. Tons of creative platforms are moving towards having generative AI labeled as such and allowing for filtering. I honestly don't understand exactly why you guys are seemingly so set in your ways with this one when it's been a very obvious complaint and issue for so long.
Just add a disclosure option for developers that says "this game contains AI generated content" and let us filter out those AI generated games if we don't want to support generative AI stuff. I really don't feel like this is a bad compromise at all, and it really is the most middle ground, easy W stance you could take and implement. Will it require effort? Yes, but it will produce a much better image for the game and people will be more willing to participate in the platform and support it. If your "main aim is that the good games float to the top and the bad games float to the bottom" then this should only help people find the games they're interested in and want to support.
I want to see this platform succeed. It's gotta feel human to succeed imo.