Epoch AI (FrontierMath / Benchmarks)https://epoch.ai/

Your landing page’s weakest score is CTA.

How Epoch AI (FrontierMath / Benchmarks) ranks

  1. 1.Epoch AI (FrontierMath / Benchmarks)76
  2. 2.ARC Prize Foundation76
  3. 3.METR73
  4. 4.Kaggle Competitions72
  5. 5.LMArena66
  6. 6.Artificial Analysis65
Fix CTA
76Growth scoreupdated

category scores

category scoresShowHide

CTA

68Fair

Copy, size, verb, cost-feel — does the primary CTA earn the click?

Epoch AI (FrontierMath / Benchmarks)
LMArenain-product action

Problems

  • 5 words — too long for one fixation
  • low-contrast mint pill, weak footprint
  • competes with '+' icons on stat cards

Information hierarchy

76Fair

Does the design help or inhibit what the page is trying to say?

Information hierarchy annotations on the desktop capture1234
  1. Truncated third stat card — The 'Training compute' card is cut off at the viewport edge with no visible caption end, signaling a carousel that hides content and adds interaction cost.
  2. Carousel arrow controls — The prev/next arrows sit far right at faint gray weight, so the paging mechanism for the stat cards is easy to miss entirely.
  3. Card plus buttons — Each card's black '+' button is the highest-contrast element in the fold, outweighing the mint primary CTA and splitting the click path three ways.
  4. Clipped section headline — The next section's headline crosses the fold half-rendered, creating a stray text fragment that competes with the logo strip.

Value props

76Fair

Title, subtitle, hero — do they sell the product's true value?

Epoch AI (FrontierMath / Benchmarks)’s headlineInvestigating the trajectory of AI for the benefit of society.

Working

Could improve

Mobile web

78Fair

The phone experience as a first-class citizen.

Epoch AI (FrontierMath / Benchmarks) mobile capture for Mobile web

What works

What to improve

Trust

82Strong

Proof that the product is real, competent, and safe to try.

Epoch AI (FrontierMath / Benchmarks) desktop capture for Trust

How Epoch AI (FrontierMath / Benchmarks) compares

Trust read