Horizon Create vs Horizon Studio: AI game creation explained
Mobile prompting and browser editing could shorten prototyping. Publishing eligibility is only the beginning of finding players.
Documentation-led analysis, not a hands-on review. Availability and announced US prices are dated; taxes, lenses and regional offers can differ. Practical examples below are proposed evaluations, not measured results.
Horizon Create is the phone workflow; Horizon Studio is the browser workflow. Meta is rolling out early access through a waitlist. Treat them as tools to evaluate for a small playable concept, not a proven replacement for your existing production process.
Source: Meta’s September 24 creation-tools announcement.
Create versus Studio
| Tool | Documented workflow | Where to start |
|---|---|---|
| Horizon Create | Mobile prompts, templates and iterative playtesting | Describe one mechanic and test whether it is enjoyable. |
| Horizon Studio | Browser prompting, visual editing, asset swaps and parameter controls | Refine a scene or rule that needs precise adjustment. |
Meta describes 2D and 3D game creation, with projects transferable between the two tools. Its developer recap says no coding is required for the intended creation flow. That means users can work without directly operating a conventional game engine; it does not mean the resulting game needs no design, testing or maintenance.
Sources: creation workflow and developer recap.
What could a creator actually use it for?
Illustrative concept: a short sorting game for a newsletter about software decisions. Players inspect fictional requirements and choose which tool fits. Each choice gets an explanation. The goal is to teach a distinction the article already makes, not to turn the newsletter into a full game studio.
A founder could similarly explore an onboarding concept; a creator could test an interactive companion to an explanation. These are proposed use cases, not claims about completed projects or conversion results.
Write the brief before the prompt
A productive first brief might read: “Build a single-player, three-round sorting game. Each round shows one fictional task and two tools. Let the player choose, show why the answer fits, then continue. No purchases, personal data or external accounts. Provide a restart button.”
This is an example brief, not a tested prompt or a guarantee of supported features. Its value is the limit it puts on the experiment. It specifies the interaction, the end condition and what is out of scope.
What still needs human control?
- The rule: can a new player explain what they are trying to do without help?
- The feedback: does the game explain a mistake clearly enough to teach something?
- The difficulty: are failures meaningful, or are they caused by unclear controls?
- The content: are names, assets and factual explanations appropriate for publication?
- The revision: does changing one requirement break a previously working round?
Generating a playable scene and producing a dependable experience are different milestones. Record the first generated result, the edits you made and the failures that remain. That evidence is more useful than a polished screenshot when deciding whether to build a second prototype.
Distribution does not guarantee an audience
Meta says published games become eligible for native distribution across Facebook, Instagram and Horizon, with review before publication. Monetization details are still forthcoming in the announcement.
Source: publishing and distribution details.
For a business, audience access should be a hypothesis you test. A game needs a reason to be opened, an understandable first interaction and a reason to return or share. Distribution eligibility by itself establishes none of those outcomes.
Availability, pricing and unanswered questions
Use the official announcement’s early-access link to check the current signup path. We have not established a public price, generation quota, full export policy, guaranteed admission date or complete commercial-use terms from the materials reviewed.
Before accepting a paid client project, verify ownership and licensing, asset provenance, supported exports, publishing review, ongoing costs and what happens if access changes. Until those are clear, scope any experiment as a prototype rather than promising a transferable finished application.
A sensible first milestone
Ask three people unfamiliar with the idea to complete the prototype without coaching. Observe where they stop, what they misunderstand and whether they can explain the intended lesson. This is a proposed qualitative check, not a statistically representative study.
Continue only if the interaction teaches or entertains more effectively than the simpler alternative you could already publish. For related tools, see the Connect guide and Muse explainer. For turning a documented evaluation into clear content, use the product-story worksheet.