4 Sources
[1]
Solo dev tries to make his own GTA 6 with AI, as he got tired of waiting for Rockstar
Thirteen years is a long time to wait for a video game. Apparently, it's long enough for someone just to say, "Fine, I'll do it myself." Ziwen Xu, a 25-year-old founder of AI agent startup Hyperecho, kicked off the project with a post on X. "Day 1 of building GTA 6. Still feels fake typing that out," he wrote, and upgraded to Claude Max 20x specifically for the project. The first footage was exactly what you'd expect: a blue box wandering around an empty 3D plane. Not much, but every game starts somewhere. He spent a week ignoring everyone who told him to switch engines before finally caving on day seven and migrating from Godot to Unreal Engine. The project, dubbed GT-Caliber, runs on daily public updates and an open GitHub repo, with AI agents continuously looping, taking community feature requests, and generating pull requests in real time. Perhaps unsurprisingly, there's no publisher, and no studio budget. Ziwen seems to be relying on whoever shows up for the project. The project has moved surprisingly fast, but also chaotically. Within the first week, Ziwen switched engines three times, finally landing on Unity. He explained that getting on a "real engine" was crucial to make a triple-A game. Of course, AI tools handled a significant portion of the asset creation, generating Rockstar-style news clips, characters, cars, and buildings. Almost everything was created through what Ziwen calls "vibe coding" with a loop of AI agents. The progress has been genuinely wild for nine days of work. By day three, NPCs were walking around, cars on the road, working weapons, and phones with Instagram in them. By day seven, a whole neighborhood with streets had appeared on screen, even though the AI accidentally built Los Angeles instead of Miami. Ziwen called it the first time the thing looked like an actual GTA game rather than a prototype. The timing of the project makes for a funny backdrop. Rockstar officially revealed the GTA 6 pre-order date today: June 25, 2026, alongside the game's official cover art. The cover follows the series' familiar collage style, featuring Jason Duval and Lucia Caminos front and center. The game still launches on November 19 on PS5 and Xbox Series X/S. So Ziwen has about five months, and his project is making progress day by day. On the flip side, Rockstar had 13 years and thousands of people working on GTA 6. Whether "GT-Caliber" ships anything resembling a real game or just becomes the most entertaining dev diary on the internet, it's worth keeping an eye on.
[2]
One solo developer got tired of waiting for GTA VI, so he decided to make one of his own with the help of an AI
Pre-orders for Rockstar's Grand Theft Auto VI start on June 25th, 2026. The actual launch is looming November 19th, 2026. A solo dev is using an AI to build his own GTA 6 clone and beat Rockstar to launch, as reported by Tweak Town. Ziwen Xu is a 25-year-old founder of an AI agent startup called Hyperecho, and he has a project like no other. He is building Grand Theft Auto VI alone, with the help of an AI. The project is called GT-Caliber, runs on daily public updates and an open GitHub repo, with AI agents continuously looping, taking community feature requests, and generating pull requests in real time. And as you might expect, he has no publisher, and no studio budget. The project has moved surprisingly fast, but also chaotically. The progress has been pretty wild for nine days of work. Pre-orders for Rockstar's Grand Theft Auto VI start on June 25th, 2026, so this solo dev's timing is interesting. It means that Ziwen Xu has about five months. What ever happens in the end, it will be an entertaining process to follow.
[3]
Dev Tries Building GTA 6 Using AI, But Accidentally Builds Wrong City Amidst Chaotic Development And Unreal Engine 5.8 Challenges
The wait for GTA 6 has been long and hard, but it is finally coming to an end. With pre-orders opening next week ahead of the game's November 19 release on PlayStation 5 and Xbox Series X|S, Take-Two Interactive and Rockstar Games are sure to ramp up marketing, and finally show more of a game set to be one of the most successful ever released. However, a solo developer has had enough, and has spent the past 12 or so days trying to build their own version of the game using AI, and the results are impressive and hilarious at the same time. As reported by TweakTown, Ziwen Xu, founder of AI agent startup Hyperecho, started working building GTA 6 using AI agents back on June 10, starting development on the Godot engine. Progress of the GT-Caliber project was incredibly fast, as by day 3, the AI agents implemented multiple community feature requests, having NPCs walking around, cars on the road, shadows and reflections working correctly. However, things started going wrong almost immediately: right from Day 2, the AI agents started building the wrong city, recreating Los Angeles instead of Miami, which is the real-world inspiration for Vice City. This AI version of GTA 6 also met a few other snags on the road. On Day 7, Ziwen Xu swapped the engine from Godot to Unreal Engine 5, while on Day 8, the developer's Mac died because every agent open the game at once to screenshot it. Still, progress continued: while NPCs lost their walking animations, the AI agents created an in-game phone with maps, contacts and a wallet. In a strange twist of fate, Rockstar Games announced the opening of GTA 6' preorders right as development of this project reached Day 9, and hit another small snag. While the map is now set and getting iterated upon, Ziwen Xu attempted to move the project to the recently released Unreal Engine 5.8 to take advantage of its Model Context Protocol, and the results weren't great, and actually slowed the project down. "The 5.8 MCP isn't like crazy good, and it actually slows the build down. Chasing the newest version felt like the obvious move. It wasn't. The loop runs better without it," the developer said. While this project is definitely fascinating, it once more confirms that AI, by itself, is nowhere near capable of developing a game like GTA 6. This is something that Take-Two CEO Strauss Zelnick is well aware of, as he believes the lack of creativity would prevent the creation of games on par with the Rockstar Games series, even if it were possible for a huge open-world game to be built autonomously by AI. Follow Wccftech on Google to get more of our news coverage in your feeds.
[4]
'GTA 6 Before GTA 6' Challenge Has an Indie Dev Racing to Finish an AI Clone Before Rockstar
GTA 6 will no doubt be the biggest game launch of 2026. With other top developers scrambling for a release date for their games that is nowhere near the GTA 6 release date, one company wants to directly challenge them instead. An AI company named Hypercho is creating a GTA 6 clone and is planning to launch it before GTA 6. Vibe Coding GTA 6 Clone is Closer to Release Than GTA 6 GT-Caliber, or the GTA 6 clone, is being developed by a small AI startup as a challenge. They are trying to see if AI can create a GTA-level game. Although the team started using Anthropic's Claude model with Godot at the beginning, they have since moved to Unreal Engine 5 to speed up the development. With fans already expecting GTA 6 trailer 3, no one planned for a GTA 6 clone trailer instead. This is the ultimate vibe coding challenge undertaken by anyone, and its scale is so bizarre that the challenge is almost laughable. However, I shudder to think what might happen if they do succeed in creating even a resemblance of a game that looks like GTA purely using AI and a couple of months of work. The only worse scenario than that is if I actually enjoy playing the GTA 6 clone after its release! Thus far, the development doesn't look anywhere near competing with Rockstar Games, and with GTA 6's budget being $3 billion, it makes sense. The graphics revealed in their latest development video show classic UE5 graphics, but the lack of style and design flatness is something you may find in a Roblox game. Although I must give credit that, in eight days, they have created a working shape of a game, and just using AI. However, even though the game is made using AI, it is not without cost. According to the developers, they are spending quite a lot on credits. This is expected, as creating a game from scratch using AI loop agents is no simple task, especially if you want the code to actually work. The community reaction has also been fairly positive so far. However, this might be because no actual gamer is taking it seriously. It may very well be seen as an AI enhancement project rather than an actual attempt at game development that is being supported here. The challenge first began on June 10, 2026. With no GTA 6 delay being reaffirmed, the game is set to release this November. So, there are still a couple of months for Hypercho to develop the game's clone. What the future holds is still unclear, but I do hope it's not filled with games made using AI. Although I personally have no issues with asset generation or R&D using AI for development, the sheer thought of having soulless games flooding the gaming market is horrifying. So, what's your opinion of GT-Caliber? Do you see yourself playing AI games in the future, or do you agree with me that the human touch in games is what makes them special? Let us know your thoughts in the comments below.
Share
Copy Link
After 13 years of waiting for GTA 6, Ziwen Xu decided to build his own version using AI agents. The 25-year-old founder of Hyperecho launched GT-Caliber on June 10, creating an open-source GTA clone with daily updates. The chaotic development saw him switch engines three times in nine days, accidentally build Los Angeles instead of Miami, and crash his Mac—but the project reveals both AI's potential and its limitations in game development.
Waiting 13 years for a video game can test anyone's patience. Ziwen Xu, a 25-year-old founder of AI agent startup Hyperecho, reached his limit and decided to build his own version of GTA 6 using AI game development techniques
1
. The project, dubbed GT-Caliber, kicked off on June 10, 2026, with a simple post on X: "Day 1 of building GTA 6. Still feels fake typing that out."1
The initial footage showed exactly what you'd expect from day one—a blue box wandering around an empty 3D plane. But what happened next demonstrates both the speed and chaos of AI agents in game development.The solo developer creates GTA clone journey has been anything but smooth. Ziwen Xu switched engines three times within the first week of development
1
. Starting with Godot engine, he spent a week ignoring advice to switch before finally migrating to Unreal Engine 5 on day seven3
. By day eight, his Mac died because every AI agent opened the game simultaneously to screenshot it3
. Eventually, he settled on Unity, explaining that getting on a "real engine" was crucial to make a AAA game1
. When he attempted to upgrade to Unreal Engine 5.8 to leverage its Model Context Protocol, the results weren't great. "The 5.8 MCP isn't like crazy good, and it actually slows the build down," Ziwen Xu admitted3
.GT-Caliber runs on daily public updates and an open GitHub repo, with AI agents continuously looping, taking community feature requests, and generating pull requests in real time
2
. There's no publisher and no studio budget—Ziwen Xu relies on whoever shows up for the project1
. Almost everything was created through what he calls "vibe coding" with a loop of AI agents1
. The AI-generated game approach handled asset creation, generating Rockstar-style news clips, characters, cars, and buildings. By day three, NPCs were walking around, cars appeared on roads, weapons functioned, and phones with Instagram were implemented1
. However, the human touch in game development became apparent when things went wrong.
Source: Beebom
In one of the more amusing mishaps of this chaotic development, the AI agents started building the wrong city almost immediately. By day two, they had recreated Los Angeles instead of Miami, which serves as the real-world inspiration for Vice City in Rockstar's GTA 6
3
. Despite this geographical confusion, by day seven, a whole neighborhood with streets had appeared on screen. Ziwen Xu called it the first time the project looked like an actual GTA game rather than a prototype1
. While NPCs lost their walking animations during the engine transitions, the AI agents managed to create an in-game phone with maps, contacts, and a wallet3
.Related Stories
The timing creates an interesting backdrop. Rockstar officially announced that GTA 6 pre-orders open on June 25, 2026, with the game launching November 19 on PlayStation 5 and Xbox Series X/S
1
. This gives Ziwen Xu about five months to complete his challenge2
. On one side, there's a solo developer with AI agents and no budget. On the other, Rockstar had 13 years, thousands of people, and a reported $3 billion budget for GTA 64
. According to Hyperecho developers, they're spending significant amounts on AI credits, as creating a game from scratch using AI loop agents is no simple task4
.
Source: Wccftech
This project confirms that AI, by itself, is nowhere near capable of developing a game like GTA 6
3
. Take-Two CEO Strauss Zelnick believes the lack of creativity would prevent the creation of games on par with the Rockstar Games series, even if it were possible for a huge open-world game to be built autonomously by AI3
. The graphics revealed in development videos show classic Unreal Engine 5 visuals, but the lack of style and design flatness resembles what you might find in a Roblox game4
. Community reaction has been fairly positive so far, though most gamers aren't taking it as a serious competitor—viewing it more as an AI enhancement project rather than actual game development4
. Whether GT-Caliber ships anything resembling a real game or becomes the most entertaining dev diary on the internet, it highlights critical questions about AI's capabilities in creating experiences that match human creativity and design sensibility.Summarized by
Navi
[2]
20 Apr 2026•Technology

26 Nov 2025•Entertainment and Society

30 Jan 2026•Business and Economy

1
Technology

2
Policy and Regulation

3
Policy and Regulation
