Six Things People Built With GPT-6 Astra: Trains, Games, And Personal Apps

TL;DR: Astra's early builders are making unusually personal things: Tom Krcha reports a steam train with 3,295 editable objects, while Pietro Schirano says he built an iPod-style Mac app for Codex threads in 15 minutes.^[1][2] These are creator reports, not controlled tests. The interesting pattern is the output: models, interfaces, and games that invite another change instead of ending at a screenshot.

Cover: LLM Rumors conceptual illustration of a maker's workbench. It does not depict the creators' actual projects; their original demonstrations are linked below.

A steam train assembled from a drawing. A portfolio folded into a virtual handheld console. A desktop companion that dresses an AI workflow in the visual language of an old music player. The first wave of GPT-6 Astra projects has a welcome sense of play.

Our previous Astra analysis focused on what happens when a model can operate software. This time, look at what people actually want it to make. The projects below range from public artifacts to creator demonstrations; they should be enjoyed with that distinction intact.

The real story isn't another race to generate the prettiest landing page. It is the possibility that more people can afford to follow a peculiar idea far enough to find out whether it works.

The Steam Train: A Drawing Becomes Material To Work With

Tom Krcha's train is a particularly good place to start. He says Astra reconstructed an old steam-train drawing in Blender as 3,295 editable objects.^[1] That number describes the creator's object count. It does not measure engineering accuracy, mesh quality, or readiness for a commercial game.

The appeal is more immediate: a reference picture becomes a scene that can be taken apart and changed. For a maker, being able to inspect the result is part of the fun. A wheel can become the next task rather than an unchangeable feature of a generated image.

Krcha's follow-up moves the idea into code. He describes two trains generated at runtime using TypeScript and Three.js, with wheel movement and explode-and-reassemble behavior.^[8] That is a different kind of creative object: a visual construction whose behavior is part of the artifact.

The practical attraction is iteration. Change the proportions, exaggerate a mechanism, or use the scene to explain how assemblies fit together. The demo is a starting point for those experiments, not evidence that technical modeling has been solved.

Watch Tom Krcha's original train demonstration.

The Personal Interface: An iPod For Your AI Work

Pietro Schirano's example brings the same spirit to everyday software. He reports building an iPod-style Mac app that visualizes Codex threads in 15 minutes.^[2] The delightful detail is the choice of metaphor. A developer workflow gets the personality of a familiar physical object.

That is a useful challenge to the assumption that productivity software must look interchangeable. A personal tool can be deliberately nostalgic, compact, or odd. It only has to make sense to the person using it.

Paul Solt offers a quieter companion example: an iOS app for weekly running mileage. He reports 13 minutes and 3 seconds for the initial demo using his AppCreator skill, then 1 hour to reach a fully functional version, with steering between build-and-run turns on his physical iPhone.^[3][10] The skill matters because the workflow includes reusable guidance, not just a model facing an empty screen.

Neither duration is a comparison benchmark. Setup, scope, revisions, and completion criteria differ. What makes both projects interesting is specificity: one person's working habits and another person's training routine become software briefs.

Watch Pietro Schirano's original Mac app demonstration.

The Playable Portfolio: Let Visitors Touch The Idea

ManuVision posted a portfolio presented as a Game Boy Advance SP-style handheld, attributing the page, 3D model, opening animation, and button wiring to Astra. The creator also links a public demo.^[4] We have not independently tested its interactions.

A portfolio normally asks visitors to scroll through evidence of someone else's work. This concept makes exploring the portfolio part of the work itself. The animation and controls have a reason to exist beyond decoration: they establish a small world with its own rules.

Here's the genius of the format: personality becomes something a visitor can do. A designer can demonstrate taste through an interaction instead of explaining it in an introduction.

That does not make the format right for every portfolio. Navigation still needs to be understandable. But it is exactly the kind of experiment that deserves to be built and tried. The creative question comes before the conversion metric.

The Exploded Car: A Demo With Somewhere To Go Next

Creator @ashebytes attributes a browser-based Model X visualization to Astra and links both a demo and source repository. The post describes an exploded view with 334 modeled pieces.^[5]

The repository credits an existing BlendKit car asset by cgi Moon. Its 334 pieces are meshes, not verified automotive components; battery, drive-unit, and suspension representations are illustrative. The achievement here is the interactive assembly and explanation, not generating a car model from nothing.^[9]

What distinguishes this example is the route beyond the social clip. Public code gives another builder somewhere to inspect the implementation and propose a change. A learning interface can evolve into a different subject, a new explanatory sequence, or a more accessible presentation.

Watch the creator's exploded-view demonstration.

The Game Studio: More Ideas You Can Actually Play

OpenAI's Playco case study supplies a commercial counterpart to these experiments. Playco says it used Astra to develop 3 themed prototypes from one grey-box foundation and required 50% fewer manual fixes than with its previous model. One cyberpunk variant still needed a performance fix.^[6]

The strategic prize is easier comparison. A team can debate a mechanic abstractly for days. A playable prototype turns that discussion into a concrete judgment: does this feel good, and what should change?

This is where whimsical community work and professional production meet. Both benefit when making an alternative becomes practical. The output creates a better conversation about the idea.

The Bigger Opportunity: Software With More Personality

The market question is whether these experiments become repeat usage. Our analytical lens puts NVIDIA and AMD on the compute side, Microsoft on developer distribution, and Alphabet and Amazon on competing platforms and cloud consumption. A compelling demo is the beginning of that demand argument, not evidence of revenue already earned.

The real story isn't a green or red price beside a launch. It is whether more people keep building, running, and revising software after the novelty fades. The watchlist above supplies a current market backdrop, not an event study: its 24-hour changes do not establish that Astra moved any of these markets. Hyperliquid's XYZ prices are perpetual-contract marks, which can differ from the underlying shares.^[11]

Let's be clear: these examples do not establish a success rate for Astra, and a launch-week selection cannot tell us how many attempts were abandoned. Creator timings are anecdotes. Public source code improves inspectability, but it does not establish authorship or production quality by itself.

The uncomfortable truth is that many good personal ideas never receive that first version. They are too small for a team, too unusual for a template, or too speculative to justify a long build. That is why a running tracker and a toy-like portfolio belong in the same conversation as a studio's prototypes.

The most exciting outcome would be a web with more personality and more people willing to build for their own curiosity. Astra's early makers offer a glimpse of that possibility. The next valuable thing may begin with a very small audience: the person who wanted it to exist.

Last updated: September 5, 2026