The free ray tracing
benchmark that runs in your browser.
A reproducible WebGL2 path tracer that measures real ray-traced GPU performance in Mrays/s. Fixed scene, fixed seed — only your hardware changes the number.
Same scene, same seed, same resolution on every device. Your score is only a function of your hardware — not your monitor or window size.
A path tracer with diffuse, metallic, and dielectric materials — running entirely in your GPU's fragment shader.
Open the tab, press run, see a number. Sign in only if you want it on the leaderboard.
What does this ray tracing benchmark actually test?
raytrac.ing renders a fixed Cornell-style scene with a real path tracer — diffuse, metallic, and dielectric materials, eight bounces per ray, accumulated across frames — and reports throughput in millions of primary rays per second. Because the scene, camera, light, materials, and PRNG seed are locked, two devices with the same GPU produce essentially the same score, and a faster GPU produces a higher one. Nothing about your monitor, window size, or browser version changes the result.
Unlike fill-rate or polygon throughput tests, a path tracer exercises the bounce-heavy, branch-heavy, math-bound side of a modern GPU — the same workload that drives real-time ray tracing in games, lookdev in VFX, and previews in 3D tooling. If you've ever wanted to know how your RTX 4090, RTX 5090, Radeon RX 9070, Arc B580, or Apple M-series chip really compares for ray-traced rendering, this benchmark tells you in about a minute.
Three sustained modes — Burst (5 s), 60 s, and 5 min — separate hot peak performance from thermal throttling, so a desktop cooler and a thin-and-light laptop with the same silicon land in different places on the leaderboard. The full method is in about.