Open it and the planet is already busy. Eleven thousand aircraft are in the air. Ships are crawling out of Rotterdam. A traffic camera in Austin refreshes, and there is a real car sitting at a real red light. Then you notice the aesthetic — tactical reticle, night-vision toggle, thermal imaging — and your brain quietly asks whether you’re allowed to be looking at this.
Quick answer: God’s Eye View is a free, open-source application that runs in your own browser, on your own machine. It doesn’t operate a satellite and it isn’t connected to any government system. It pulls publicly broadcast data — aircraft transponders, ship beacons, orbital catalogues, seismic feeds, city traffic cameras — and paints all of it onto a photorealistic 3D Earth at once. Most of it works with no account and no payment. The spy-movie interface is a design choice; the data underneath is ordinary open data, assembled unusually well.
It was released under the MIT licence in late August 2026 by Bilawal Sidhu, a former Google product manager who worked on AR/VR and 3D Maps and now hosts technology coverage for TED. Within weeks it had become one of the most-starred new repositories on GitHub.
What actually happened, and why it blew up
The project existed before the code did. Sidhu had been posting “god view” demos on YouTube under the name WorldView, and they travelled — the series has passed 5 million views on YouTube and 25 million across social platforms. Viewers kept asking for the source. He polled them, they said yes, and he shipped it. GitHub
The response was disproportionate even by viral-repo standards. It reached number one on GitHub Trending on both the daily and weekly boards in August 2026, landed at #8 Product of the Day on Product Hunt after being posted by Chris Messina — the person who proposed the Twitter hashtag — and drew a short, approving note from Brendan Eich, the creator of JavaScript and co-founder of both Mozilla and Brave. GitHub
As of mid-September 2026 the repository sits at roughly 36,000 stars and 7,200 forks, with more than a hundred open pull requests. That growth curve is the story: this stopped being one person’s demo and became a community project in under a month.
One small correction to how this often gets retold online: Eich co-founded Mozilla, the organisation behind Firefox. He did not personally create Firefox, and the distinction is worth keeping straight.
What it is — and what it definitely isn’t
It helps to be blunt about the category. This is a data-fusion front end. It is not imagery from a satellite you control, and there is no live video of the Earth.
Under the hood it’s deliberately unglamorous: vanilla JavaScript, CesiumJS for the globe, and Vite for the build, with Google Photorealistic 3D Tiles supplying the planet and the OpenAI Realtime API powering voice. No framework. The whole point is that you can read it. GitHub
The repository’s own disclaimer is more candid than most commercial products manage. It states plainly that the data may be delayed, incomplete, modelled, inferred or simply wrong, and that the tool should not be used for flight or maritime navigation, emergency response, medical decisions or anything else safety-critical. Take that seriously. It is an exploration toy and a learning instrument, not an operational picture. GitHub
How to install God’s Eye View
There are two supported paths, and the first one requires no coding at all.
Path 1 — one click, no terminal. Install Pinokio, a desktop launcher for local AI and web apps, then open the featured God’s Eye View entry and press Install, then Start. It handles dependencies, finds a free local port and opens the app.
There’s a catch worth knowing if you tried this earlier and failed. The original launcher tripped an install bug inside Pinokio itself. Pinokio’s maintainer fixed it in version 8.2, released on 3 September 2026, and re-tested the launcher across operating systems before featuring it. If your install broke before that date, update and try again rather than debugging your own machine. Pinokio
Path 2 — terminal or coding agent. Clone the repo, run npm ci, then npm run doctor and npm run dev, and open http://localhost:4173. It needs Node.js 24.14.0 or later, or 26.x; the setup doctor will flag Node 25, which has reached end of life. GitHub
Both paths are available on Windows, macOS and Linux. And both start without any API keys at all — you get Esri satellite imagery in 2D, with OpenStreetMap as a fallback, and most of the live layers work immediately.
What it costs, honestly
This is where most coverage gets vague, so here are the real mechanics.
Keys are upgrades, not requirements. You add them inside the app through a panel the interface calls POWER UP, not by editing config files. Six providers are supported, four of them have free tiers, and the layers that need nothing at all include flight data, satellites, earthquakes, public cameras, radio and launch schedules. GitHub
For the photorealistic 3D globe, the cheap route is a free Cesium ion token. Cesium’s Community plan is free for individuals and explicitly limited to personal, non-commercial use, and it includes 1,000 Google Photorealistic 3D Tiles root tile requests per month. The paid alternative is a direct Google Maps Platform key, which also unlocks in-app place search but requires billing to be enabled. cesium
Three more free-signup keys fill out the map: AISStream for ships, NASA FIRMS for active fires, and TomTom for live traffic flow speeds.
The only thing that reliably costs money is voice. Realtime audio runs to a few cents per active minute, and the app meters it for you: a live session-spend readout sits beside the microphone, with a $2 warning and a $5 hard cap that terminates the session. That’s an unusually responsible piece of design in a category where runaway API bills are a standing joke. GitHub
The data is real, but it isn’t coming from space
Here’s the part the spy-thriller framing obscures, and it’s the most interesting thing about the project.
The aircraft are not observed by satellite. They’re broadcasting. Modern aircraft transmit position, altitude and identity over ADS-B on 1090 MHz, and anyone with a cheap receiver can hear them. Those receivers belong to hobbyists.
The OpenSky Network, one of the feeds God’s Eye View uses, is a Swiss non-profit that collects this data from volunteers around the world who run receivers and push the results to a central server. It began in 2013 with eight sensors in Switzerland and Germany and had grown past 6,000 registered receivers worldwide by 2023. oxox
So: a distributed, volunteer-built, continuously updated digital twin of global air traffic, maintained by strangers with antennas on their roofs. The source video described this as “tens of thousands of volunteers.” The verifiable figure is thousands, which is remarkable enough without inflation.
Why the Strait of Hormuz looks empty
Now the lesson that makes this tool genuinely educational.
Turn on vessels and Europe lights up. Rotterdam is dense, the English Channel is a traffic jam. Then fly to the Strait of Hormuz — one of the most strategically watched waterways on Earth — and it’s close to blank.
That isn’t censorship. Ships broadcast AIS, the maritime equivalent of ADS-B, but the signal is low to the water and needs a receiver within roughly line-of-sight range of the coast. Where volunteers and port infrastructure are dense, coverage is excellent. Where they aren’t, the ocean goes quiet.
The fix is satellite-based AIS, and that’s exactly where free open data ends. The project’s own documentation is direct about this ceiling: terrestrial AIS goes silent mid-ocean, satellite AIS costs real money, and premium imagery, SAR and the deeper commercial feeds sit behind enterprise contracts. GitHub
Learn that once and you’ll never read an OSINT map the same way again. Empty space on a live data map usually means “no sensor,” not “nothing happening.” That distinction is the difference between analysis and self-deception, and it’s why this tool is more useful as a teacher than as an intelligence product.
Talking to the planet
Voice mode is the feature that makes demos go viral, and it’s more than a fancy remote control.
The agent runs on twenty-eight tools and pulls live scene context before answering, including coordinates, street names, active layers and view scale — so asking “what city is this?” mid-flight actually works. You can ask it to fly somewhere, orbit a target, draw a route between two points and then fly that route, or outline a real administrative boundary as a persistent annotation. GitHub
At street level it will read a screenshot of the viewport to identify legible signage and building names, and it’s instructed not to invent labels it can’t see. That constraint matters more than it sounds. The failure mode for an AI narrating a map is confident fabrication, and designing against it is the right call. GitHub
One honest limitation: the world knowledge it offers about a base, a port or a power grid comes from a general-purpose language model, not from the map data. It’s context, not verification. Treat it as a well-read companion rather than a source.
What people are actually building with it
The most substantive use so far isn’t spy cosplay. It’s disaster reconstruction.
On 26 August 2026, a mass of ice and rock broke away from the Langtang Himal in Nepal’s Rasuwa district and mobilised into a catastrophic debris flow down the Bhote Koshi and Trishuli rivers. The USGS catalogued the event as a landslide rather than an earthquake, recording a seismic signal equivalent to magnitude 5.2, with a total runout close to 100 kilometres. By mid-September the death toll had reached 1,396, with more than 5,100 people still missing. gee-community-catalogground
That is the event people have been rebuilding inside God’s Eye View — plotting the flood path against the terrain and pinning the videos and photographs that surfaced along its route. It’s possible because the underlying data was released openly: Vantor (formerly Maxar) activated its Open Data Program on 27 August at the request of the Humanitarian OpenStreetMap Team, and Planet Labs tasked collections through its Disaster Data programme. gee-community-catalog
Point an AI coding agent at the repository, describe the layer you want, and you can add it. That’s the actual product here: not fifteen data layers, but a working pattern for a sixteenth.
The surveillance question deserves a straight answer
A visible share of the reaction to this project has been alarm rather than delight — concerns about privacy, public cameras, national security and what happens when many separate public feeds are fused into one operational picture. Some people have argued it shouldn’t be publicly available at all.
That concern is reasonable, and it’s worth engaging with rather than dismissing. Aggregation genuinely changes things: a hundred harmless public feeds combined can reveal patterns no single feed does.
The project draws an explicit line. Its documentation states that it models events, assets, infrastructure and systems — aircraft, vessels, satellites, fires, cameras, cities — and that it will not build features for named-person search, face recognition or individual tracking, with pull requests crossing that line refused. The README’s phrasing is blunt: “People are not a query type here.” GitHub
The hardest test of that policy so far is the automated licence plate reader layer, which the community requested and the maintainers merged. It maps camera locations drawn from OpenStreetMap’s crowdsourced DeFlock tagging. It shows manufacturer and operator tags where they exist, flags cached or incomplete coverage, and does not display camera footage or plate records. GitHub
That’s a defensible reading of the line: it maps the surveillance infrastructure watching the public, not the public. Whether that line holds as thousands of contributors push layers is an open question, and it’s the thing worth watching.
There’s also a practical risk that has nothing to do with ethics. The app binds to localhost by default, and for good reason. If you expose it on your local network, it brokers your configured API keys to anyone who can reach it. Don’t share an instance casually. GitHub
What’s next: a hosted version is confirmed
The single most-requested feature after launch wasn’t a new layer. It was a link.
The maintainers have now committed to it. The plan was originally to keep the repository as an open-source client and build a separate professional product; after launch, the team confirmed in the README that an official hosted God’s Eye View is being built at Halfpixel, requiring no installation. The project is maintained by Bilawal Sidhu and Sameh Khamis at Halfpixel. GitHubGitHub
No date, no pricing and no feature list have been published, so treat the timeline as genuinely unknown. The commercial obstacle is real: several of the free data feeds are licensed for personal, non-commercial use only, and a free public website still counts as commercial under many of those terms. Solving that means swapping providers or paying for them.
The other stated ambition is harder still. The maintainers note that the present is the cheap part — going back in time, and tiling and serving what changed at real resolution, is where the data gets expensive and the compute gets brutal. That’s an accurate description of why historical geospatial products cost what they cost, and a reasonable expectation-setter for anyone hoping for a free scrubbable archive of the planet. GitHub
The honest verdict
If you want an operational intelligence tool, this isn’t it, and its own documentation says so first.
If you want to understand how the modern world is actually observed — who broadcasts what, where the sensors are, where the gaps are and why they exist — it’s the most effective teaching instrument available for free right now. The tactical HUD is a costume. The education underneath is real.
Start keyless. Turn on flights. Then go look at the Strait of Hormuz and work out why it’s empty.
FAQ
Is God’s Eye View free?
Yes, and the code is MIT-licensed. It runs with no account and no keys. Free-tier signups unlock ships, fires and live traffic. The only feature that reliably costs money is AI voice, which is metered and capped at $5 per session.
Is God’s Eye View a real satellite?
No. It shows no live satellite video. It renders publicly broadcast data — aircraft transponders, ship beacons, orbital catalogues, seismic feeds and city cameras — over pre-existing 3D map imagery.
Is it legal to use?
It uses publicly published data feeds, each with its own terms. The most common restriction is non-commercial use, which applies to the free Cesium ion tier and several data layers. Personal exploration is within terms; redeploying it as a public service is not, without changing providers.
Why can’t I see any ships in some regions?
Ship AIS signals are received by coastal receivers. Where there’s no nearby receiver coverage, ships don’t appear even though they’re there. Global mid-ocean coverage requires paid satellite AIS.
Do I need to be a programmer to install it?
No. The Pinokio launcher is a two-click install on Windows, macOS and Linux. If an earlier attempt failed, update to Pinokio 8.2 or later first.
Is there a website version I can just open?
Not yet. A hosted version has been confirmed by the maintainers but has no announced release date or pricing.
How is this different from Flightradar24 or MarineTraffic?
Those are polished single-domain trackers. God’s Eye View fuses many domains into one 3D scene and lets you extend it with your own data layers, at the cost of being rougher and less reliable than any of them individually.
