What's My AIMeasured browser confidence matrix, refreshed from cohort telemetry.
GuideMeasured cohorts

guide

Browser support now follows measured cohort evidence instead of static browser copy.

The matrix below groups completed benchmark cohorts by browser, browser version, device class, GPU family, and accelerator path. Confidence drops automatically when a cohort goes stale or stays on a fallback path.

measured cohorts2
tracked score version2026.03-browser-v6
latest refreshOct 29, 2026
next step
Run the benchmarkRead the FAQ

embedded benchmark pilot

Need the answer fast? Open the full benchmark from this guide.

This compact entry keeps the browser-support guide as the attribution source, then opens the full benchmark on WhatsMy.AI in the top window when someone decides to run it.

Chrome

Measured coverage. 1 completed runs across 1 cohort.

Latest browser version seen. 155.0.0.0.

Effective confidence band. low.

Measured accelerator paths. Full WebGPU.

Reason codes. Full WebGPU cohort, Thin coverage, Primary browser path.

Firefox

Measured coverage. 1 completed runs across 1 cohort.

Latest browser version seen. 140.0.

Effective confidence band. low.

Measured accelerator paths. WebGL-only.

Reason codes. WebGL-only cohort, Thin coverage, Stale cohort, Fallback browser path.

Measured cohort matrix

Browser cohortConfidenceEvidence
Chrome 155.0.0.0 • desktop • Apple M2 • Full WebGPUlow (68%)1 runs • Full WebGPU cohort, Thin coverage, Primary browser path
Firefox 140.0 • desktop • Intel(R) HD Graphics, or similar • WebGL-onlylow (0%)1 runs • WebGL-only cohort, Thin coverage, Stale cohort, Fallback browser path

How the fallback works

Full WebGPU cohorts can earn the highest confidence only after repeated measured runs land on the strongest accelerator path.

Degraded WebGPU, WebGL-only, and no-accelerator cohorts stay capped lower even when the raw score looks good.

Any cohort that stops refreshing inside the weekly window is automatically downgraded until newer runs arrive.