When Apple Vision Pro launched in February 2024, the headline take was "too expensive, not enough games, niche product." Eighteen months later that framing doesn't quite hold. Apple shipped the second-generation Vision Pro M5 in late 2025, the Vision Air (the cheaper headset, $1,799) in March 2026, and visionOS 3 opened real APIs for spatial persistence, multi-user shared spaces, and direct GPU access that the original launch firmware lacked. The install base is still small compared to Quest — roughly 2.5 million units across both generations as of April 2026 — but the per-user spending is not small, and the indie success stories are real.
This post covers what's actually different about building for visionOS, the engines and tools that work in 2026, the App Store economics that make small-team spatial games viable, the design patterns that earn good reviews, and where the platform's limitations still break indie projects. If you're weighing whether to port an existing game or start a visionOS-native project, this is the honest April 2026 landscape.
The State of the Platform
Apple Vision Pro M5 is a $3,499 spatial computer with micro-OLED displays at roughly 23 megapixels per eye. Vision Air at $1,799 trades the high-end displays for pentile OLEDs (still sharp, not quite best-in-class), reduces battery life from 2.5 to 2 hours, and drops some hand-tracking sensors while keeping full feature parity in APIs. Both run visionOS 3.
The install base breakdown as of April 2026:
- Vision Pro (gen 1 + M5): ~1.4M units
- Vision Air: ~1.1M and growing faster than Pro did
- US-heavy but now available in 22 countries
The key takeaway: this is not a mass-market platform like Quest 3, but the owners are disproportionately high-income, software-engaged, and willing to pay for quality apps. App Store reports suggest the median spatial-app user spends 3-5x more than a median iPhone user.
What's Different About visionOS
visionOS is not iOS with a VR skin. The meaningful differences that matter to indie devs:
Mixed reality by default. Most spatial apps are MR, not pure VR. Your game runs in the user's actual room, windowed or immersive, and the room's lighting, surfaces, and geometry are available to your app through RoomPlan and SceneKit's mesh reconstruction. "Pure VR" immersive mode exists but is not the default expectation, and many successful spatial games blend MR gameplay into real space.
Eye and hand are the primary inputs. No controllers as first-class citizens. You can pair a PS5 or Xbox controller for games that need them, but the platform-native input model is: look at a thing, pinch to select. This is genuinely different from Quest and forces rethinking input design from scratch.
Shared space and Full Space. Apps can run in "shared space" (alongside other apps, in windowed volumes) or take over "full space" (exclusive access, immersive). Games mostly want full space. Productivity-hybrid apps live in shared space.
Spatial Personas and multi-user. visionOS 3 added Shared Space sessions where multiple Vision Pros can share a spatial experience — think local co-op in a real living room, not matchmade online. This is a genuinely new gameplay surface.
No background processing for games. Games don't run when the user looks away. Expect frequent state save/restore, no push notifications, no background music while doing other things.
Engines in April 2026
Unity 7 with PolySpatial 3 is the primary cross-platform path. PolySpatial now compiles Unity projects to visionOS's RealityKit runtime with most of URP's feature set working. The gaps: custom shaders still require rewriting in MaterialX, some particle systems behave differently, and Unity UI needs fully rebuilt for gaze-and-pinch. Budget 3-6 weeks for an experienced Unity developer to port a simple existing game.
Unreal Engine 5.7 has experimental visionOS support through the "Apple Spatial" plugin. It works for immersive VR-style games but not for windowed MR experiences. Epic has been noncommittal about full PolySpatial-equivalent support; expect clarity in UE 5.8 or UE 6.
RealityKit + Swift is Apple's native path. Best integration, lowest friction, highest ceiling on quality. The tradeoff: you're writing Swift and there's no cross-platform story. For visionOS-exclusive games, especially small-scope ones, this is actually the right choice. Reality Composer Pro is a surprisingly good 3D authoring tool once you learn its conventions.
Godot 4.6 has community-maintained visionOS export in alpha. It technically works for simple games. It is not yet production-ready and Apple's TestFlight certification process occasionally rejects Godot builds for missing entitlements.
Bevy and other Rust engines: no viable visionOS path in April 2026.
What Actually Sells on the Platform
The spatial App Store in April 2026 skews toward these game genres:
Room-scale puzzle games. Puzzling Places, Super Fruit Ninja Pro, and dozens of indie follow-ups show that "pick up and manipulate spatial objects in your room" is the genre the hardware most enables. Tight scope, 8-20 hour experiences, $15-25 price points.
Meditative / atmospheric experiences. Journey-to-the-sacred-foundation of visionOS. Slow-paced, visually lush, short sessions. These sell steadily to the "I just want to decompress" audience.
Cozy MR games. Virtual pets that live on your desk, gardening games that grow plants on your windowsill, tiny cities that sit on your coffee table. This is the genre most suited to Vision Pro's form factor and most overserved on Quest.
Cinematic light-gun / wave shooters. The "stand still and shoot" genre maps well to the hardware, especially for Vision Pro's older demographic.
Board game and tabletop conversions. A virtual table in your dining room running Catan or Terraforming Mars is shockingly good on visionOS. Digital conversions of premium board games are a reliable seller.
What doesn't work: fast-paced competitive shooters, any game requiring sustained head movement for 30+ minutes, games that assume twin-stick controllers as the native input, games that need huge play areas.
The Design Patterns That Earn Reviews
Spatial game reviews are disproportionately harsh on a handful of sins. Avoid these and your reviews will be a standard deviation better:
- Don't make users stand if they don't have to. Most visionOS sessions are seated. Design for seated play first, standing as optional.
- Respect the room. If the user has a chair behind them, don't spawn enemies there. Use RoomPlan data.
- Keep sessions 10-30 minutes. The headset gets heavy, eyes get tired, batteries die. Design for short sessions and natural stopping points.
- Audio is primary, not secondary. Spatial audio is a signature platform feature. Games that lean into 3D audio review noticeably better.
- Motion sickness is still real. Any artificial locomotion without strong comfort options will hurt your reviews.
- Don't require the user to turn around. Their couch has a back. Design for a roughly 180-degree front-facing play space.
App Store Economics
The economics that make visionOS attractive for small teams:
- The store is uncrowded. ~3,500 native spatial apps in April 2026 vs 1.8M on iOS. Being found is dramatically easier.
- Paid apps work. Unlike iOS (where free + IAP dominates), paid visionOS games at $10-30 sell reasonably well. The demographic has paid for a $1,800+ device; $20 is not a barrier.
- Apple features indies. The "Discover" tab on visionOS App Store has aggressive indie curation. Quality projects get featured.
- 30% / 15% standard Apple cuts apply. Small Business Program gets you to 15% on first $1M.
Reports from shipping indies suggest a well-received $20 visionOS game can generate $50K-200K in first-year revenue on modest marketing. That's not category-defining money, but for a 3-6 month project from a small team, it is viable — and the catalog of shipped spatial indies keeps generating long-tail revenue because the store isn't flooded.
When visionOS Is Wrong for Your Project
- Your game needs fast reflexes and controller precision
- Your target audience is under 25 (install base skews older)
- Your game assumes 2+ hour continuous sessions
- You have no budget for the 3-6 week visionOS-specific polish pass
- You don't own a Vision Pro — you genuinely need the device to test
The last point is underrated. Developers building for visionOS without access to a headset ship broken games. Simulator testing is insufficient for anything beyond basic layout.
A Practical Port Decision
If you have an existing game, the port calculus:
- Existing Unity 3D puzzle/cozy/narrative game: probably worth porting. Budget 6-10 weeks.
- Existing Unreal AAA-style 3D game: probably not worth porting yet. Wait for UE's story to solidify.
- Existing 2D game: only worth it if the game has a natural spatial reimagining. Not most 2D games.
- Existing Quest game: often worth it. Quest and Vision Pro audiences overlap less than you'd think, and the same build works differently on each platform.
The Honest Outlook
Vision Air's price drop was the structural event that made visionOS a platform indies can target, not just a tech demo. The install base will cross 5M by late 2026 on current trajectory. The App Store is still uncrowded enough that good indie games get discovered. The economics are comparable to launch-era iPad in 2010, which in retrospect was an excellent time to be an indie.
Not every indie should build for visionOS. But "spatial computing is a 2030 problem" is no longer true. For the right kind of small, focused game, April 2026 is a defensible time to start.
Related reading: VR Game Development for Meta Quest in UE5, Indie Game Discoverability 2026, and Micro-Studio Tech Stack 2026.