Adrián

Here’s my submission for the #vibejam @levelsio @cursor_ai @tripoai contest: AniKombat. AniKombat is a browser-based 3D fighting game built with Three.js. It has combos, instant VS CPU matches, training mode, casual online multiplayer, and ranked ladder play. No login or signup is required to play. You can jump straight into Training, VS CPU, or casual online matches as a guest. Accounts are optional and only used for ranked progression, match history, and ladder climbing. I built this in about 2-3 weeks using AI tools: - @cursor_ai and @OpenAI with GPT-5.4/5 for coding - @threejs for the 3D game client - @tripoai for 3D model assets - @ElevenLabs for sound effects/music - GPT Image for artwork, UI direction, and visual planning . This was a great learning experience. The hardest part was definitely online networking: sync, interpolation, prediction, reconciliation, and making combat feel playable over latency.

@nypost Notice how these so-called journalists frame things.. 🤡 “For a deadly strike in the occupied east.” Conveniently, you forgot to mention WHY. This was a retaliatory strike after Ukraine carried out a terrorist attack that killed innocent Russian children. x.com/rt_com/status/…

Russia launched a major ballistic missile assault on Kyiv late Saturday, carrying out Moscow’s explicit vow to punish Ukraine for a deadly strike in the occupied east.

@triggerfxx @hqmank No. It’s good - very good, actually, compared to most other models, but it’s nowhere near GPT-5.5 level. I have run a lot of side-by-side tests with both, and 5.5 consistently performs WAY BETTER.

@hqmank Is V4 Pro really on the same level as 5.5 high? Like genuine question.

Codex is currently the best AI product.🔥

@Laraesku Convenientemente, olvidaste mencionar el por qué. Esto fue un ataque de represalia después de que Ucrania cometiera un acto terrorista que mató a niños inocentes en Rusia. Dobles estándares de mierda. x.com/rt_com/status/…

🎙️Han sido 5 horas de intensa ofensiva. Rusia ha lanzado 600 drones y 90 misiles, que han tenido como objetivo principal Kyiv. Los ciudadanos se refugiaron en los andenes del metro, a decenas de metros de profundidad, porque cerca de 40 localizaciones estaban siendo atacadas: viviendas, oficinas, o un mercado. Al menos dos personas han muerto y la cifra de heridos aumenta conforme pasan las horas: ya unas 80 personas. Se registraron múltiples incendios simultáneos, por toda la ciudad. Y durante las labores de limpieza y rescate, Emergencias alertó a residentes de bloques afectados que no se aproximaran a los fragmentos de los drones kamikaze, porque Ucrania ha detectado que, en ataques recientes, partes de dron albergaban componentes radiactivos inestables. Y otro dato: al menos uno de los misiles que logró impactar esta noche fue el hipersónico de medio alcance Oreshnik. Un arma difícil de interceptar, con capacidad para transportar ojivas tanto nucleares como convencionales. Sería la tercera vez que Rusia lo emplea sobre Ucrania. @noticias_cuatro #ucrania #kyiv #київ #noticiasinternacionales #UkraineRussiaWar️

@Breaking911 Conveniently, you forgot to mention WHY. This was a retaliatory strike after Ukraine committed a terrorist act that killed innocent children in Russia. x.com/rt_com/status/…

Russia unleashed one of its largest attacks on Kyiv since the war began, hammering the capital and nearby areas with hundreds of drones and missiles, including the Oreshnik hypersonic missile.

@thsottiaux Mine went from 85% to 25% after one prompt... 😟

Seeing issues where usage limits are out of sync for some Codex users. Apologies and team is investigating.

@thsottiaux Yes Tibo

Codex usage limits have now been reset across all paid plans. Enjoy the weekend!

@GraemeVIP @ajambrosino I have a couple of Mac minis at home that run my dev environments 24/7. Honestly, the real perk is being able to monitor or manage things even if I’m out of the house. It’s mostly just about flexibility/convenience

Thanks for the feedback on Codex in the ChatGPT mobile app. While it’s in preview, we’re working to improve it fast. What you can expect next: push notifications, /fork, ability to restore after revoking, better reconnects, fixing the ability to control other devices, fewer mobile thread errors, better git diff & full-file, no plan mode issues, and lots more polish/bug fixes.

@trekedge Can ya wait like 1 more hour pretty plz 🥺 I still got like 60% left to milk 🥲

Use fast mode tonight reset incoming

@gmcharmander @thsottiaux dm sent 👍

@adrscott @thsottiaux hi adrián, yes please, that would be great if you can share the session ids for these!

Codex team is aware of reports of GPT-5.5 performing worse for some users and investigating. We don't have anything conclusive yet and systems are healthy but we will share updates as we go.

It’s also been acting really off and unusual. Almost like it’s “forgetting” things or repeating the same mistakes even after I reminded it beforehand. I’ve got a few /feedback examples I can share if that helps. Last night alone I wasted like 3 hours rebuilding a C++ project because it kept touching .h files it shouldn’t have. I complained about it, and it still kept doing it… @gmcharmander @thsottiaux

@thsottiaux Don’t care about a reset (still ~60% left until the 19th), but the past few hours with 5.5-High have been really bad. A /goal trashed my codebase (thank god for Git) wasting hours and ~20% of my quota. Took a break because I just don’t trust it right now. It’s performing rly bad.

@NicholasRuncie @thsottiaux Same experience here. I asked it to fix something that was mostly just config tuning. Took a break, came back, and it had started changing core code instead lol. It’s been acting really unusual.

@thsottiaux Has this been resolved yet? Something still feels off. Asked codex to solve a bug and it ended up deleting features rather than solving the root cause.

@jacklovesfreddy @yacineMTB That’s why you run regression tests and use git.

@yacineMTB And then you wake up and the whole code base is broken and you hit your limit

/goal optimize this code, use `date` to check the time and don't stop until I wake up at 7 am

@thsottiaux Did you guys nerf 5.5? Today it’s been really bad. It keeps making mistakes, getting stuck in loops, and failing to fix bugs it introduced itself. It fees like a different model 🤔

You can just migrate things

Your first preview of Mortal Kombat II Release date: May 8, 2026 Recored by me 😁 at cinema

@DevanshArya12 @sama Mostly yep 😁

@adrscott @sama Oh man, this is all vibe coded 😳!!??

i would like to talk to people who have built amazing things with 5.5 that weren't possible with earlier models. i am especially interested in examples that took ludicrous token budgets. thanks.

@aniketmandle11 @sama Three.js. The scenes are all built/rendered in WebGL with custom assets + animations. GPT-5.5 helped wire the gameplay logic, camera, controls, combat states, etc.

@adrscott @sama How are the characters and scenes rendered ?

Imagine spending all day crying about AI on X instead of actually building something 😭 AI didn’t magically solve networking sync, interpolation, prediction, reconciliation, combat design, or make the game fun to play. I still had to build and ship it. I used the tools available, learned a ton, and made a multiplayer 3D fighting game for a game jam. Stay mad I guess 🫵😂

@adrscott @sama The Ai made it, you didn’t. And it looks horrible. Why not learn how to make actual games?

@Ruan3D_ @sama It was for the vibejam contest! I had a lot of fun making this game :) x.com/adrscott/statu…

@adrscott @sama Looks like something from 2001. What’s the use case?