GLB Optimizer

Compress and optimize GLB models

Shrink GLB and glTF files for the web and AR — Meshopt or Draco geometry compression, WebP textures, polygon reduction and texture downscaling, all in your browser. Free, private, and nothing is uploaded.

GLB / glTF inputDrop your GLB or glTF file hereor click to select it (add textures/.bin for glTF). Files never leave your device.

Max texture sizeDownscaling oversized textures is usually the biggest size win.

Weld vertices

Merge duplicate vertices. Lossless — keeps hard edges and UVs.

Reduce polygons0%

gltfpack-grade quadric simplification — preserves UV seams & hard edges. Skips rigged & morph meshes.

Geometry compressionMeshopt decodes quickly into GPU-ready buffers. Draco often compresses geometry further and has broad viewer support, but decodes more slowly.

Textures to WebP

Re-encodes color textures when smaller. Data maps keep their original format; selected downscaling still applies.

Capabilities

What the optimizer does

Choose Meshopt for fast, GPU-ready web delivery or Draco for the smallest geometry buffers.
Re-encode color textures to WebP only where it genuinely shrinks them — material data maps keep their original format.
Downscale oversized textures without changing their aspect ratio (4K → 2K/1K/512) — often the single biggest size win.
Reduce polygons with meshoptimizer’s seam-aware quadric simplifier — the same engine as gltfpack.
Weld vertices bit-for-bit, prune only safe unused data, and preserve every scene, animation, skin, morph target and named hierarchy node.
Reads Draco-, Meshopt- and KTX2-compressed inputs. 100% in-browser — your model never leaves your device.

Process

How it works

Drop your GLBDrag a .glb or .gltf file (add its textures/.bin for glTF) onto the tool. It loads locally in your browser.
Pick your settingsChoose a max texture size and geometry codec, keep exact vertex welding on, and optionally dial in polygon reduction.
Optimize & downloadSee exactly how much smaller it got, then download the optimized GLB — ready for the web or AR.

Notes

Good to know

Meshopt and Draco output require a decoder in the destination viewer. Choose “Off — widest compatibility” for older tools.
Both geometry codecs use controlled attribute quantization; choose Off when exact floating-point vertex data is required.
Polygon reduction is lossy by nature; rigged and morph-target meshes are kept at full detail to protect animation.
KTX2 textures pass through unchanged; this browser tool does not transcode or resize GPU-compressed texture payloads.
Models using unsupported custom glTF extensions are rejected instead of silently losing extension data.
Very large scenes are processed on your device, so speed depends on your machine.

FAQ

Questions, answered

Is the GLB optimizer free?

Yes. Optimizing and downloading models is completely free — no account or payment required.

Are my files uploaded to a server?

No. Everything runs locally in your browser. Your model never leaves your device.

How do I make my GLB smaller for AR?

Use Draco when your AR viewer supports it, keep WebP textures on, and set a max texture size of 2048 or 1024 px. Preview the result before shipping because polygon and texture reduction are intentionally lossy.

Can it optimize Draco-compressed GLB files?

Yes. Draco- and Meshopt-compressed inputs are decoded automatically, optimized, and written with whichever output codec you select.

Will optimizing break my materials or animation?

The direct glTF pipeline preserves scenes, materials, skeletal animation and morph targets. Polygon reduction deliberately skips rigged and morph meshes, while color WebP conversion and texture downscaling remain optional.

Keep going

USDZ optimizerShrink textures without rebuilding USD STL optimizerClean and simplify triangle meshes GLB editorFull browser-based GLB workflow All convertersGLB, OBJ, STL, FBX and more