How was the 144fps / 12,000-node benchmark measured?

The test runs a scene-graph stress workload of 12,000 concurrent animated DOM nodes on a Pixel 6 (mid-tier Android reference device) for 60 seconds. Sustained 144fps means the 95th-percentile frame interval stayed under 6.94ms with zero dropped frames over the full run. The full harness, device list, and audit logs are published in the docs.

Which devices were included in the test matrix?

Five tiers: low-end Android (Samsung A13), mid-tier Android (Pixel 6), high-end Android (Galaxy S23), mid-range laptop (2020 Intel i5 / integrated GPU), and M-series MacBook Pro. The 144fps result holds across all five tiers; full fps and jitter numbers are tabulated on this page.

Can I reproduce the benchmark myself?

Yes. The reproducible harness, scene fixtures, and CLI runner are published in the CodeAnimato docs under /benchmarks/reproduce. The docs page is the primary call-to-action from this report.

Performance dossier · v4.0 "Volta"

144fps is not a marketing line. It is a reproducible result.

CodeAnimato's headline number — sustained 144 frames per second across 12,000 concurrent DOM nodes on mid-tier Android — was measured, audited, and published in the 2024 WebPerf Report. This page is the test matrix behind that claim, so you can verify it before you sign up.

Read the docs v4.0 release notes

Methodology audited quarterly · Trusted by 41,300+ developers across 162 countries

scene.timeline.ts

// CodeAnimato 4.0 — scene graph declaration
import { Scene, Layer, spring } from '@codeanimato/core';

export const hero = Scene('hero')
  .mount('#root')
  .fps(144)
  .nodes(12_000)
  .add(
    Layer.grid(100, 120)
      .ease(spring({ stiffness: 180, damping: 22 }))
      .commit()
  );

avg 142.7fps · 12,000 nodes · 14.6KB gzip

§ 01 — Methodology

How the test was run.

Every figure on this page comes from a public, versioned harness — not a cherry-picked demo. We publish the device matrix, the frame-budget methodology, and the audit cadence so a skeptical senior engineer can reproduce the run in an afternoon. The full harness lives in the docs; the summary is below.

Device matrix

Twelve reference devices spanning five tiers, from a 2019 Pixel 3a (Snapdragon 670, 4GB RAM) to an M3 Max MacBook Pro. Mid-tier Android — the actual differentiator against lighter competitors — is represented by five handsets including the Samsung A52 and the Xiaomi Redmi Note 11. Each device runs the harness at native resolution with no thermal throttling mitigation beyond a 90-second cooldown between runs.

Frame-budget methodology

We measure the 99th-percentile frame time over a 60-second continuous animation, then convert to sustained fps. Jank is defined as any frame exceeding the 6.94ms budget required for 144Hz. A run is reported only if zero dropped frames appear in the final 10 seconds — warm-up frames are discarded. Battery, GPU driver version, and Chrome flags are recorded per run.

Audit cadence

The full matrix is re-run on every minor release and again at the end of each quarter. The Q1 2024 audit produced the median bundle size of 14.6KB gzipped across customer builds. Independent replications are welcome — the harness, the seed scenes, and the raw CSV exports are linked from the docs page.

§ 02 — Results readout

The headline numbers, on a single page.

A flat readout. No spinners, no animated counters — the data is the headline.

144 fps

Sustained frame rate at the 99th percentile across the 12,000-node scene, on every device in the matrix.

12,000 nodes

Concurrent animated DOM nodes driven by a single scene graph — the load that defines the claim.

14.6 KB gzip

Median runtime bundle across audited customer builds, Q1 2024. Hard cap stated in the v4.0 release notes: 18KB.

99.99% SLA

Enterprise tier uptime, with a sub-50ms timeline-jitter guarantee for hosted timelines.

Adoption 41,300+ developers

Reach 162 countries

Survey 4.7× YoY — Stack Overflow 2024

Audit Quarterly · public harness

§ 03 — Device coverage

Same scene. Five tiers. Locate your target hardware.

The 144fps result holds across the matrix — including a 2019 mid-tier Android — not only on flagship laptops. The table below is excerpted from the Q1 2024 audit; the full per-device CSV is in the docs.

data-label="Tier" = Mid-tier laptop

Tier	Reference device	Resolution	Sustained fps	Jank (frames > 6.94ms)	Runtime
Low-end mobile	Pixel 3a (2019)	1080 × 2160	144	0.0%	13.9KB gzip
Mid-tier Android	Samsung A52	1080 × 2400	144	0.0%	14.6KB gzip
Mid-tier Android	Xiaomi Redmi Note 11	1080 × 2400	144	0.1%	14.6KB gzip
High-end mobile	iPhone 15 Pro	1179 × 2556	144	0.0%	14.8KB gzip
Desktop (flagship)	M3 Max MacBook Pro	3456 × 2234	144	0.0%	15.1KB gzip
Low-end desktop	Dell XPS 13 (i5-1135G7)	1920 × 1200	144	0.2%	14.7KB gzip

All runs use the same scene-graph declaration shown in the hero panel. WebGPU timeline scrubbing is enabled only on devices where the API is available; results above were collected with WebGPU disabled to keep the comparison fair across tiers.

§ 04 — Independent verdict

"CodeAnimato is the only animation runtime we measured at sustained 144fps across 12,000 concurrent DOM nodes on mid-tier Android. It is, in our reading, the current benchmark for production-grade web motion."

2024 WebPerf Report Independent performance audit · 47 frameworks measured

Featured in the 2024 Stack Overflow Developer Survey CSS Design Awards — Tool of the Year, 2024 #1 Product of the Day, Product Hunt (April 2024)

§ 05 — Reproduce it yourself

The full harness lives in the docs.

Clone the benchmark repo, pick a target device from the matrix, and run the harness locally. Every figure on this page is reproducible from the same scene-graph declaration — no special build flags, no insider telemetry.

Read the docs See what changed in v4.0

Public roadmap with 87% of shipped features originally proposed by community users
Open-source core — 38.2K GitHub stars — under MIT
Optional paid Studio and CI plugins for teams that need more