The Sandbox has wrapped three and a half months of closed testing on its AI game-creation platform, The Sandbox Studio — and more than 100 games came out of it. The company confirmed a public launch in October, opening the door for anyone to build games with AI tools inside one of crypto’s best-known metaverse brands.
By Imani Davis | October 4, 2026
The Hook: From 12,000 Applications to 100-Plus Games
In an October 3 post on X, The Sandbox summarized the testing window: eight product releases, four game jams, more than 100 handpicked creators, and more than 100 games built. Partners and integrations named include Meshy and Scenario, two tools that turn text prompts into 3D and visual assets, with the company promising “more soon.”
The journey to get here was deliberately exclusive at first. The closed alpha drew more than 12,000 applications, but only about 30 creators got in during the first phase, with The Sandbox saying it would admit up to 10 more creators each week. The creators selected had experience with Roblox, Unity and Unreal Engine, and many had already used AI coding assistants such as OpenAI Codex. The beta opened on August 12 to anyone who had applied for the alpha.
How the Games Were Built: AI Under the Hood
The Sandbox Studio is a bet that ordinary people — not just professional studios — can make real games if AI does the heavy lifting. Think of it as a kitchen where the AI is the sous-chef: you describe the dish, and the tools handle the chopping. The company has called the tech stack open, listing Codex, Claude Code and Gemini among the options, and said from the platform’s announcement in June that creators could bring their own AI tools, naming Cursor, Claude Code and OpenAI Codex.
- 100-plus games — built during 3.5 months of closed alpha and beta testing
- Eight product releases — shipped across the testing window
- Four game jams — including a Beta Creator Game Jam with a 7,500 USD prize pool
- Meshy and Scenario — AI asset tools that convert prompts into rigged, textured 3D character models, listed alongside Tripo on the creator hub
- Web, mobile and Telegram — the distribution channels, with Steam, mobile app stores and YouTube Playables planned over time
The Core Conflict: AI Speed Meets Crypto Distribution
For NFT watchers, the interesting tension is what Studio means for The Sandbox’s original model. The platform made its name on user-owned land and tradable digital collectibles. Studio leans on a different engine: AI-generated content at scale, distributed far beyond crypto-native channels — including Telegram Mini Apps, which puts games in front of the messaging app’s massive user base.
The game jams showed how the community responded. The first jam, open only to alpha creators, ran July 2 to 31 with a 5,000 USD prize pool split among the top five submissions and a 2,500 USD top prize, with winners announced August 7. The Beta Creator Game Jam carried a 7,500 USD pool under the theme “Overgrown,” judged by Larry Hryb — the man who spent 15 years as the public face of Xbox. A Halloween Game Jam is now listed as live on the creator hub, and in August The Sandbox added a section to its site that features games built with Studio.
During the alpha, the team also laid groundwork for multiplayer, leaderboard services, asset collections, character rigging and better support for Agent Nova, the AI assistant built into Studio — a checklist that reads like a real game platform rather than a crypto marketing stunt.
Market Implications: What This Means for You
If you hold SAND, The Sandbox’s token, or NFTs tied to its ecosystem, the public launch is the number to watch. Closed tests with 100-plus creators are one thing; opening the doors to everyone in October is when the platform’s content pipeline either floods or trickles. More games means more players, more activity, and — in the platform’s thesis — more demand for the digital assets attached to it.
For the broader NFT market, this is a template worth noting. NFT trading volumes have cooled dramatically from their peaks, and the projects still standing are the ones pivoting from speculation to actual utility: games people play, tools people use. An AI game builder that ships a hundred titles before launch day is exactly that kind of pivot — a shift from “buy this JPEG” to “make this game.”
The Verdict
The Sandbox Studio’s closed testing phase delivered a concrete result: a backlog of more than 100 games and a public launch date this month. The waitlist at create.sandbox.game is open, and the real test begins when the public gets its hands on the tools. If the quality holds up beyond the handpicked creators, Studio could become one of the largest experiments yet in blending AI creation with blockchain ownership. If it does not, it will be another promising beta that never became a business. October will start answering that question.
The cryptocurrency market remains highly volatile. This article is for informational purposes only and does not constitute financial advice.
12,000 applications and they let in about 30 creators for phase one. hope the october public launch actually has capacity for everyone else
tbf closed alphas always start small, the 12k applications number is just marketing framing. still waiting to see one of these ai built games thats actually fun to play
jamestether_ the four game jams were basically public playtests then. 100 games shipped and nobody can name one, thats the review
phase one was about a hundred handpicked creators per the post, not 30. either way 12,000 applications says demand is not the bottleneck, capacity is
capacity wont be the problem, curation will. 12k applications crashing into an open editor means the catalog turns to slop by november
slop by november maybe, but the current closed catalog is already 60 percent asset flips and it still ships. curation is a store problem, the editor being open just moves the noise earlier
curation being a store problem is generous, the jams had no store at all. the october launch ships with charts day one or the slop argument never ends
roughly 100 games from roughly 100 creators in 3.5 months. about 3 games per creator, either the ai tooling is real or the counters are generous
100 games from roughly 100 handpicked creators in three and a half months. The quiet story here is the tooling, Meshy and Scenario for asset generation is where the speedup actually lives.
Kwame A. meshy for 3d and scenario for the skins is half a pipeline already. the missing half is gameplay loops and ai still cant write those
Kwame A. meshy and scenario cover art but nobody ships a game on assets alone. the eight product releases are the real pipeline story, they were patching editor gaps all test long
Meshy doing the 3d and Scenario the skins is exactly where the speedup lives. asset work was always the week-long part, code is the fast bit now
belinda nailed it, the meshy pipeline is the whole story here. my last jam game was 90 percent asset work, cut that to a weekend and the 100 games counter is a floor not a ceiling
12k applications and 30 creators in phase one, then they act surprised the open launch needs a waitlist. the exclusivity was the marketing and it worked on me, i already signed up
signed up too and i know exactly zero people who got rejected from the waitlist. curious what the october capacity actually is, the 12k number set expectations rough
Same here, signed up day one and nobody I know got rejected either. Either the october capacity is way bigger than they let on or the waitlist is theater before the open launch
a weekend each with ai art help and still nobody outside the jams can name one of the 100. the october launch answer is retention numbers, not build counts
^ this. 100 built means nothing until one of them shows up on a telegram chart. hope the october launch surfaces a single actual hit
100 games in 3.5 months sounds wild until you remember jam games are a weekend of work each with ai art help. the october public launch is when the numbers get honest
honest metric is day 7 players not games built. if the october public launch ships a retention graph even once ill be shocked
day 7 is the right metric. jam games live and die in 48 hours, so if a few of the 100 kept players past week one that is the whole story
one honest retention graph from the october launch is worth more than every build counter combined. publish d7 or it never happened
agreed, one retention graph ends this argument. if the october launch shows daily actives per game the builder count stops mattering overnight
eight product releases in one beta tells me they were patching the editor against real complaints, not the roadmap. that matters more than the 100 games counter
eight releases also means eight rounds of jam feedback each time. betas that patch weekly are the only ones that survive contact with 12k randoms in october, so ill take the churn over a quiet roadmap