How WebGL Reports GPU Info
WebGL itself only reports a generic vendor/renderer string like "WebKit" or "Google Inc." by default - it does not identify your actual graphics hardware. The real vendor and renderer strings (for example, "Intel Inc." and "ANGLE (NVIDIA, NVIDIA GeForce RTX ...)") are only available through the optional WEBGL_debug_renderer_info extension, which a page must explicitly request with gl.getExtension('WEBGL_debug_renderer_info') and which the browser is free to withhold.
Beyond identity, a WebGL context exposes a large set of numeric capability limits through gl.getParameter() - things like MAX_TEXTURE_SIZE, MAX_VIEWPORT_DIMS, and MAX_CUBE_MAP_TEXTURE_SIZE - plus a list of every supported extension via gl.getSupportedExtensions(). This tool reads all of that after creating a context, trying webgl2 first and falling back to webgl.
Why Vendor/Renderer Gets Masked
An exact GPU model string is a strong fingerprinting signal - combined with other browser characteristics, it can help make an individual browser session more uniquely identifiable across sites. Firefox disables the debug renderer info extension by default in its standard privacy settings, and Safari has historically restricted or generalized it too. Chrome and Chromium-based browsers (Edge, Opera, Brave) generally still expose it by default, though this can change with browser updates or privacy-hardening extensions.
This tool detects that gap directly: if the extension request succeeds, it shows the real strings; if it comes back generic or fails, it shows what was actually returned along with a clear note explaining that the browser is masking it - rather than presenting a masked value as if it were a genuine one.
Common Use Cases
Reading the Capabilities
| Field | Source | Meaning |
|---|---|---|
| Unmasked Renderer/Vendor | WEBGL_debug_renderer_info | Real GPU identity, when the browser allows it |
| Max Texture Size | MAX_TEXTURE_SIZE | Largest single texture dimension the GPU supports |
| Max Viewport Dims | MAX_VIEWPORT_DIMS | Largest renderable viewport size in pixels |
| Extensions | getSupportedExtensions() | Every optional WebGL feature this context supports |
| Logical CPU Cores | navigator.hardwareConcurrency | Number of logical processors available to the browser |
Frequently Asked Questions
On Windows, Chrome and Edge typically render WebGL through ANGLE, a layer that translates OpenGL ES calls to Direct3D. That's why an unmasked renderer string often looks like "ANGLE (NVIDIA, NVIDIA GeForce RTX 3060 Direct3D11 vs_5_0 ps_5_0, D3D11)" rather than a plain GPU name - it's still identifying your real GPU, just wrapped in ANGLE's naming format.
No. navigator.deviceMemory is deliberately coarsened to a small set of values (such as 0.25, 0.5, 1, 2, 4, or 8 GB) and is capped at 8 even on machines with far more installed RAM, specifically so it can't be used as a precise fingerprinting signal. It's also Chromium-only - Firefox and Safari don't implement it at all.
No. It's a hard capability ceiling reported by the driver, but allocating a texture at that size can still fail at runtime if the GPU doesn't have enough free video memory available. The value is a limit, not a memory or performance guarantee.
Not necessarily. getSupportedExtensions() only reports that the browser/driver claims to implement an extension, not that every code path in it is bug-free or equally fast on a given GPU - some extensions have historically had inconsistent driver-level behavior across vendors.
The browser's own WebGL implementation stack (such as ANGLE), the OS graphics API it targets, and browser-specific extension allowlists all sit between the raw GPU and what JavaScript sees. Two browsers running on identical hardware can genuinely report different limits or extension lists as a result.