Sneeze: An Open Metaverse Browser Engine, and the JanusVR Idea Standardised
Published: June 15, 2026
TL;DR: At AWE 2026 in Long Beach, the Metaverse Standards Forum introduced Sneeze — billed as the world's first open metaverse browser engine (MBE). Today's engines — Blink in Chrome, WebKit in Safari, Gecko in Firefox — were built for 2D documents. Sneeze is built for spatial computing: self-hosted "spatial fabrics" — explicitly the metaverse equivalent of websites — that you run on your own infrastructure and navigate between. If that sounds familiar, it should: it's the JanusVR model, now formalised into an open standard. It's open source under Apache 2.0, and it's credible day-one infrastructure to watch — not, yet, a finished product.
What was announced
Sneeze is the first project out of the Open Metaverse Browser Initiative (OMBI), a working group inside the Metaverse Standards Forum created together with RP1. It landed on 15 June 2026 at AWE and shipped the same day as open source — Apache 2.0, on GitHub, with documentation at omb.wiki.
The argument is simple. The browser engines we all use were designed around the page — a 2D document you scroll. Sneeze is purpose-built for spatial content instead, and is meant either to be embedded into an existing browser or used to build standalone native metaverse browsers from scratch. The announcement carries endorsements from Neil Trevett — president of both the Khronos Group and the Metaverse Standards Forum — and Sean Mann, co-founder and CEO of RP1, which is acting as lead architect. The University of Rochester used the moment to launch an Open Metaverse Academic Alliance (OMAA) alongside it.
What Sneeze actually does
| Capability | What it means |
|---|---|
| Spatial fabrics | Self-hosted 3D spaces, explicitly described as the metaverse equivalent of websites — you run them on your own infrastructure, with no central platform in the middle |
| Proximity-based discovery | Content loads as you move through a physical space; the airports, hospitals and factories cited in the announcement are the example use cases |
| Scene Object Model (SOM) | Composes one 3D scene from multiple origins — different servers each contributing to the same space |
| Per-service WASM sandboxing | Each service runs isolated inside WebAssembly, so one origin can't reach into another |
| Real-time co-presence | Shared, live spaces across AR glasses, VR headsets, phones and desktops |
Built on standards, not reinvented
The most reassuring thing about Sneeze is how little of it is new. It leans on the boring, proven plumbing of the web — HTTPS, TCP/IP, DNS — and on existing open specifications: Khronos's ANARI, OpenXR, SPIR-V and glTF, plus W3C's WebAssembly and Decentralized Identifiers. The genuinely new piece is the spatial composition layer — the Scene Object Model and the engine itself. For the technically curious, the repo lays out the stack:
| Concern | Technology |
|---|---|
| Core | C++ static library |
| Rendering | ANARI, via a Filament-based device called Halogen |
| WebAssembly | Wasmtime |
| XR | OpenXR |
| UI | HTML/CSS, rendered with RmlUi |
RP1 is building the first native browser on Sneeze — the repo references an app called Artemis — with more promised at AWE and in the weeks that follow.
We've seen this idea before — it was called JanusVR
Here's the part that made us sit up. Linked, self-hosted 3D rooms that you walk between, each one served from your own infrastructure like a web page — that is not a new idea. It is, almost exactly, the model JanusVR and its FireBoxRoom format pioneered back in 2013–14. VR Sites was one of the first platforms hosting JanusVR rooms in those days, so we're not guessing at the resemblance — we ran it.
The rhymes are hard to miss. "Spatial fabric = website" is the old JanusVR pitch almost word for word. Self-hosting a room on your own server was the entire point of FireBoxRoom. Even the detail of describing the interface in HTML and CSS — Sneeze does its UI with RmlUi — echoes how Janus rooms were authored. To be clear: there is no stated connection between the two projects, and we're not claiming one. But the shape of the thing is unmistakable.
The difference — and it's the one that matters — is who is building it. A decade ago this was a scrappy community idea with a couple of clients and a handful of room hosts. Now a recognised standards body, with Khronos and RP1 behind it, has built infrastructure for what that early WebXR/Janus community was already doing by hand. The idea didn't change. The backing did.
Keeping it honest: This is credible infrastructure from a real standards body with named backers — but it is not a shipping product. The public repository is currently a fork with minimal traction and no tagged releases, and the first browser built on it is still being written. Treat Sneeze as early-stage infrastructure to watch, not a finished thing you can hand to users today. A day-one announcement is a starting line, not a finish.
The bottom line: Sneeze takes the most interesting idea the early immersive web ever had — linked, self-hosted 3D spaces that behave like websites — and gives it the two things it never had the first time around: an open standard and serious institutional weight. Whether it actually ships is still an open question. But that the standards world has finally formalised the JanusVR model is, from where we're sitting, the real news.
Sources & further reading:
Metaverse Standards Forum — Sneeze announcement ·
GitHub — MetaverseStandards/Sneeze ·
omb.wiki ·
omb.wiki/sneeze
Hosting your own corner of the immersive web — then or now? Share your self-hosted 3D space with us at VRSites.com. We've been cataloguing linked, browser-based VR since the Janus days, and we're still at it.