Tone

I have finished reviewing/getting impressions of every single game out of the 945 games submitted to #vibejam. If you submitted a game and haven't received any feedback or tags from me, take a look at the link below. Impressions tend to be brief. To be frank, some of these games did not deserve even the brief time I put into them. But SOME of these games are genuinely stunners. Some of these games have enormous depth, gorgeous visuals, addicting loops, and genuine commercial potential. I'll be posting some of my thoughts, opinions, and wrapup ideas as the official judging of the vibejam continues. As a reminder, I am not an official vibejam judge. I just wanted to look at all of the competition and see what the playing field looks like.

@dangreenheck I think there was a massive surge of excitement around the 2026 vibejam, but that's over

Is it just my feed or did the whole vibe-coding thing kind of die off?

@MrForExample @fightlifemobile

🥊5 months into development Motion fluidity still not quit there yet, but at least head movement, hands movement and footwork are all procedurally generated by the system in real time according to player input, and they are all works together now, as you can see in the video below when control our guy with keyboard & mouse Next, will add physics component, so this guy can finally hit something #gamedev #indiegame #boxing

@realBigBrainAI Why not post the link and credit the creator instead of making this sound like some spontaneous generation of anons? capybara-vibejam26.leocoout.dev @leocooout

This is what vibe coding actually looks like in 2026: a capybara food delivery game, built solo in 14 days with zero game dev experience.

@leocooout Would love to read that breakdown. Is this still on your radar or did you finish it and I missed it?

i wanna do a tech breakdown of how and what i've built to support the development of my game capybara-vibejam26.leocoout.dev

decided to search about my vibejam game and it's crazy the amount of people doubting it took me 2 weeks to deploy a workable mvp with assets, music and code made with AI FOR A JAM

@JohnOcens @UsuallyTheHero @RagimMusakaev @sebastien_and @fabiensnauwaert It’s pretty magical isn’t it? All these people with creation now at their fingertips. Best. Possible. Timeline.

It really was an experience creating this. It's the first time in a long time I've felt proud to reach the end. And playing all the other games and thinking how many others would have felt the same way made me feel that the future will be beautiful. My favourites are hoverx and neon runner - their creators were attentive to the parts of the experience that make games thrilling.

#vibejam quick reviews 011-015 1. Sock Climber @UsuallyTheHero 2. Almost Surgery @RagimMusakaev 3. Clawthrone @sebastien_and 4. LingoGoat @fabiensnauwaert 5. Mazima @johnocens Worth a play? ⬇️

I can't help but wonder if, in the same way you transitioned from creator to supervisor, if this isn't just a facsimile of the age old process of reaching as far as you can (and would like to) in your current role. But when working with agentic coding, there isn't really a role above supervision of the work. There is no human worker's needs to consider, no human decisions on benefits that need balancing against human shareholders and stakeholders, no politics to wrestle or intractable morons to work around. The problems lose some aspect of human dimensionality, so what then is the next role? My hunch is it needs to involve people. But perhaps you're already doing this?

Thread. I thought I was immune from ever feeling hollowed out by AI as a programmer, because I've always gotten far more enjoyment from shipping, getting users, and solving problems than indulging in the art of coding. As the LLMs have eaten deeper and deeper into our field, I've empathized with my peers who've expressed a sense of loss and disillusionment as the art of programming has become more and more automated. But, I've always seen myself as someone who saw coding as a means to an end to solve problems. Not something whose craftsmanship, culture, methodologies, and fads were worth getting too hung up on, beyond what was needed to adeptly deliver value to others and not fall behind the (frankly, rare) genuine advancements over the years. This all changed for me over the last week. The frontier probably shifted a bit earlier than today, but I didn't see it until now. The change has come about for me because GPT-5.5 was able to build complex software I needed built autonomously for 2-3 days at a clip. Work that would have taken me months, or even years if you include learning the requisite languages, libraries, and tooling, being completed over a weekend. This isn't something I think anyone who has been programming as long as I have can really be prepared for, this kind of velocity jump is just mindboggling. This is truly superhuman performance - it's not perfect, and there certainly is a level of simplicity and clarity that would come in the hands of the world's best programmers, but that margin is so small so as to be unnoticable when contrasted with the sheer volume of working software that it can produce per unit time. So, why has this caused a shift in the way I feel about these technologies, after all this time not having felt it as each subsequent model advanced closer to what we see now? There are two reasons. First, it's clear that the age of humans understanding how software works is over. Yes, humans will need to understand things, at least for a few more years, but we are now at a kind of escape velocity where the % of lines of code that are created every year that are even read, nevermind understood, by humans, is now permanently declining. But the real shift, is I am no longer a programmer, I am a manager. Good managers do not take credit for the work of their team - they see themselves in service of their team. Up until now, claiming "I built this" still felt true when talking about things I had created with the help of LLMs. But now, when the LLMs are writing thousands of lines of code, and I am simply providing guidance, direction setting, and oversight to catching the bigger errors, I found myself in the bizarre situation (that many will be in soon, I presume) of no longer feeling entitled to take credit for the work being done. Not being able to say "I built this" when sharing something whose basic conception came from my own mind, but under the tireless effort of these insane machines to actually reason through and materialize into a working solution, is devastating. Not because of the fact it doesn't feel truthful now, but because I know it will never be truthful again for myself and soon for all of the rest of us.

@JohnOcens @UsuallyTheHero @RagimMusakaev @sebastien_and @fabiensnauwaert Glad to hear it was more than just some prompts tossed out and abandoned for you! Keep building

Thank you for the review, Tone. I created Mazima because I believe all the experiences I've ever had in games were far more than the game itself; it was about the people I spoke to or played with, even if I never know their names. In the background, where you see the glowing orbs travelling, they each represent another player in the world moving alongst the valley with you in realtime. The world of Mazima is a circular valley you travel all the way around with other players. They are all around you; you just can't see them.

@georgebsocial Cap is a good program. But also gifs are good for scrolling - they just play

It's 2026. #indiedev's should stop showing their games with gifs. Record videos. nVidia makes this quite easy. So does OBS. I'm sure AMD cards have some way to record video. Preferably with sound if you can and it's not too early. #gamedev Also 60fps video please. No more gifs! They do your game a disservice.

@gfodor Insane. OG

@InOneProjects This didn’t actually hit me personally until I got to 8 hour trajectories, after 3-4 years now of using LLMs at the frontier to build software. I think it’s going to hit more devs soon when they get where I am now.

Comprehension debt isn’t just a liability to teams because of the obvious problem of loss of human oversight. The bigger problem is the fact that you can’t starve engineer-minded people of the joys of true understanding for long before morale suddenly collapses.

@DougTenNapel Love it. “I don’t build in order to have clients. I have clients in order to build.”

You can’t create if you obey an audience. Just give up the creation now because you’ve lost the charter before you even started. I care about the thing I’m making so much that I even have to mitigate my own voice sometimes. I’ll ruin the thing even as it’s trying to be something true good and beautiful. How much less the audience and commerce matters at the start of a story. After it’s done the audience, marketing, economy and critics are allowed to have a say. You get to respond to what we make after it’s done.

My friend, that’s probably thousands of lines of code. You’ll be reviewing for the next few days and at some point you’ll have the “doing this all at once was not cost or time saving - it just moved all the review into a huge hard to move chunk” revelation if you haven’t already 😂 good luck though

@InOneProjects Oh this is all the stuff that has no "feeling" as such It's a shed load of refactoring, DX improvements and potential fixes. (to be confirmed) You underestimate my capacity to review code 😉 I've seen done some chunky pull requests in my career!!

I may have let ultracode run all night (On the Trizen engine) with commit perms... It effectively did thermo-nuclear-code-review and then self applied fixes. That'll be a fun PR to look at after work 🫠 At a glance I can see some stuff that definitely needed sorting!

@NoemiTitarenco You’re saying “env” and not “E N V”?

I also use voice-to-text for my prompts and I need HELP. How can I say "ENV" without the voice-to-text thinking I said "M"?

Great take! “It’s AI slop!” Is just a teenager girl social dismissal attack. To other teenagers, it is a pronouncement of reality imposed socially. To adults who can think it’s nonsense.

@kunchenguid @GergelyOrosz Good take

i think any hope of a “fire and forget” solution is misguided. this can’t be solved with a one-size fits-all approach that gets implemented top down it needs to be done by each engineer, or each team to figure out what’s the sweet spot for their work. this is no different from other engineering efficiency efforts - there’s no magic button that says “make our entire company’s CI pipeline 5x faster”. if there was, it would have been enabled by default on the vendor side the solution i suggest is to have different tiers of token limit everyone starts with a generous enough tier that gives them enough to experiment and learn but only people who have demonstrated strong ROI get promoted to a higher tier this will incentivize employees to figure out how to achieve better efficiency in their environment, which scales much better over the long term

So, at your company, what are you doing about the spiking token prices? Approaches I hear: 1. Inference routing - route more basic workloads to a LOT cheaper APIs or providers 2. Use cheaper open models for inference (eg Fireworks, Baseten etc) 3. Default model: cheap one

I misread @mitchellh as Michelin and read this entire thing through the lens of "Michelin star restaurant vs not" expecting it to make comparisions between optimizations in food and code (it doesn't) and the point still stands 😅 Go to any casual restaurant and try their food. If they could optimize their food so it was even twice as good - and it was easy - would this not be a huge win? Would you not pick mom and pop burger over Red Robin (never again!) if their food was better, but still not Michelin star? This is the huge gap in everything software surrounding AI. The galaxy brains (who earned their galaxy brains) are stuck in the "Everything must be Michelin star code - what will you do when there's a bug you can't solve!" mindset and they miss a sea of small improvements, decent experiences you never heard of, games that never would have been made, and who knows, maybe the occasional absolute connoisseur who does a great job with simple tools.

@Daniel_Farinax Ah the FAQ I was looking for!

Only the Grok Imagine API can do this! I built a fully automated pipeline that creates cinematic videos for my entire game X-Quadron programmatically, at scale. Dozens of custom videos. Zero manual editing. Every scene now feels like a movie. How I did it: Went to Grok API from xAI docs site Copied the video generation docs Told my AI agent to build the whole flow That’s literally it. Here’s the full breakdown + how the videos look inside the actual vibe-coded game. What would you build with this? Drop your wildest idea below 👇 Follow for more AI-powered game dev magic.

@aipulseda1ly What's the prompt? Wouldn't mind trying on my end

Opus 4.8 comparison vs GPT-5.5 Pro and Standard, NYC skyline prompt, same task Both Opus 4.8 results are the best I have ever seen on this prompt. Nothing else comes close Opus 4.8 in Claude Code with max thinking took 40 minutes and produced a result more detail than anything I have tested. The model thinks roughly 10x more than 4.7 before it builds Opus 4.8 on the web UI is a different story. Thinking so long it hits the message length limit, forces a continue, breaks mid-thought, then finishes with "lost memory" warnings. Still the best visual result I have seen on this prompt but the web experience is broken right now Claude Code is the right environment for this model. The website is not ready for what 4.8 actually does

These are such cool benchmark prompts. I’m going to start doing the same thing each time a model comes out. Nothing sells “this is different” like a powerful visualization