How fast does the banner appear?
Three in four visits showed it within 26 ms.
All 10 CMPs, best first
- Inth25 ms
- Cookie Control375 ms
- CookieYes393 ms
- iubenda409 ms
- Enzuzo449 ms
- Didomi456 ms
- Osano625 ms
- OneTrust640 ms
- Usercentrics704 ms
- Ketch1.82 s
In our tests, Inth showed its cookie banner in a median 25 ms on desktop. It delayed the main content by 16 ms (95% interval 4 ms to 24 ms). Accept responded in 40 ms and Reject in 40 ms. It held trackers back in all 4 consent checks.
On an emulated mid-tier phone, the banner took 252 ms. Per vendor docs, Inth supports IAB TCF, IAB GPP, Google Consent Mode v2 and Global Privacy Control, but not Google CMP certification.
Three in four visits showed it within 26 ms.
Compared with the same page without a consent manager.
Reject responded in 40 ms. Under 200 ms feels instant.
No tracker ran and the page set no cookie before the visitor chose.
Each status comes from Inth’s documentation. CookieBench does not test these features unless a line says so.
| Feature | Status |
|---|---|
| IAB TCF | Per vendor docs (vendor documentation, opens in a new tab) |
| IAB GPP | GPP 1.1 (vendor documentation, opens in a new tab) |
| Google Consent Mode v2 | Per vendor docs (vendor documentation, opens in a new tab) |
| Google-certified CMP | Not on Google’s list (vendor documentation, opens in a new tab) |
| Global Privacy Control | Per vendor docs (vendor documentation, opens in a new tab) |
| Feature | Status |
|---|---|
| Geo-targeted banners | Per vendor docs (vendor documentation, opens in a new tab) |
| Script blocking | Per vendor docs (vendor documentation, opens in a new tab) |
| Consent records | With a backend (vendor documentation, opens in a new tab) |
| Cookie scanning | Via Inth Website Audit (vendor documentation, opens in a new tab) |
| Feature | Status |
|---|---|
| Open source | Apache-2.0 (vendor documentation, opens in a new tab) |
| Self-hostable | Per vendor docs (vendor documentation, opens in a new tab) |
| Headless UI | Per vendor docs (vendor documentation, opens in a new tab) |
| TypeScript types | Per vendor docs (vendor documentation, opens in a new tab) |
| Feature | Status |
|---|---|
| Free tier | Open source; hosted free plan (vendor documentation, opens in a new tab) |
Support per Inth’s documentation.
Each row is a separate app. We measured each one against the same app without a consent manager. The rows compare integrations, not frameworks.
| Framework | Banner appears | Page slowed by | Click responds in | Trackers blocked |
|---|---|---|---|---|
| Script tag | 288 ms | +4 ms95% interval 4 ms to 16 ms | 48 msOne visit measured | 4 of 4 passed |
| React | 273 ms | +20 ms95% interval 16 ms to 20 ms | 48 msOne visit measured | 4 of 4 passed |
| Next.jsShared with every CMPShown above | 25 ms | +16 ms95% interval 4 ms to 24 ms | 40 msOne visit measured | 4 of 4 passed |
| Nuxt | 16 ms | No measurable delay95% interval 0 ms to 28 ms | 40 msOne visit measured | 4 of 4 passed |
| Astro | 14 ms | No measurable delay95% interval 0 ms to 20 ms | 32 msOne visit measured | 4 of 4 passed |
| TanStack Start | 24 ms | No measurable delay95% interval −4 ms to 24 ms | 32 msOne visit measured | 4 of 4 passed |
Lab detail for engineers. All Core Web Vitals are here. The summary leaves some of them out.
The banner was visible at 25 ms. The published results do not show when Inth started to load.
| Event | Time from navigation |
|---|---|
| First byte (TTFB) | 8 ms |
| Banner visible | 25 ms |
| First paint (FCP) | 40 ms |
| Consent state known | 76 ms |
| Largest paint (LCP) | 92 ms |
| CMP requested | |
| CMP loaded |
Each row shows the value on the page with the consent manager. It also shows how far the value moved from the same page without one.
| Measure | With Inth | Difference |
|---|---|---|
| TTFBTime to first byte | 8 ms | +5 msNo interval reported |
| FCPFirst Contentful Paint | 40 ms | +12 msNo interval reported |
| LCPLargest Contentful Paint | 92 ms | +16 ms95% interval 4 ms to 24 ms |
| CLSCumulative Layout Shift | 0.00 | No measurable difference95% interval: 0.00 |
| Primary content paintThe page’s main content, not the banner | 92 ms | +16 ms95% interval 4 ms to 24 ms |
LCP times the largest element on screen. If that is the banner, LCP measures the banner instead of the page.
The screen responded to Accept in 40 ms.
| Event | Time from the click |
|---|---|
| Screen responds | 40 ms |
| Banner gone | |
| Consent applied | |
| Analytics active |
Initial transfer happens before the page settles. Deferred transfer comes after.
| Measure | With Inth | Difference |
|---|---|---|
| Initial transferBytes before the page settles | 717 KB | +104 KB95% interval: 104 KB |
| Deferred transferBytes loaded later | Not compared | |
| RequestsAll requests | 16 | +6No interval reported |
| Third-party requestsRequests to other hosts | 0 | 0No interval reported |
Long tasks are stretches over 50 ms when the page cannot respond to taps or keys.
| Measure | With Inth | Difference |
|---|---|---|
| Long-task timeBlocking time during load | 0 ms | 0 msNo interval reported |
A visitor who accepted on an earlier visit comes back. We check that the banner stays hidden.
matrix-2026-10-15On desktop, it appeared in a median 25 ms, against 452 ms for the median of the 10 CMPs measured. On an emulated mid-tier phone, it appeared in a median 252 ms, against 2.61 s for the median of the 10 CMPs measured.
On desktop it delayed the main content by 16 ms (95% interval 4 ms to 24 ms). On an emulated mid-tier phone it delayed the main content by 400 ms (95% interval 396 ms to 408 ms). Delay is the extra time before the main content appears. We compare Largest Contentful Paint on the same page, with and without a consent manager.
It passed 4 of 4 checks on desktop and on an emulated mid-tier phone. A check fails if a tracker runs, the page sets a cookie or data leaves the page before the visitor chooses.
Per vendor docs, Inth supports IAB TCF, IAB GPP, Google Consent Mode v2 and Global Privacy Control, but not Google CMP certification. Feature support comes from Inth’s documentation. CookieBench did not test these features itself.