Ranked by what each change is worth on the composite — the gap to the next rung of the ladder times the dimension’s weight. Computed from the scores, not written by the model; the model’s sentence under each is the action.
If aiming higher, reduce abstractions elsewhere (e.g., avoid words like 'workspace' in the supporting copy).
Follow with a clearer CTA like 'Install now' or 'Get started now' tied to the product claim.
Include a customer name with numbers or a verifiable usage metric.
All three together: 73 → 81. Every dimension one rung up: 81 (grade A). Ceiling is 95.
Built only from facts already on the page — not the company’s words. Shown so the gap between what was written and what could have been written is concrete rather than a number.
Five alternative headlines in five angles, a subhead, a call-to-action line, and a line-by-line edit for each of the five dimensions — written only from facts already on supercomputer.sh, gated like a rating (anything with a struck word is dropped, not sold). Delivered on a private page in about a minute.
Five dimensions on one ladder — 5 · 20 · 35 · 50 · 65 · 80 · 95 — weighted clarity 30%, specificity 20%, jargon 20%, proof 15%, call to action 15%. For each: the rung the page matched, the words that drove it, and what would reach the next rung. Full rubric →
The headline uses a concrete product noun (GPUs) and a clear target (agents).
GPUs for agents, loved by humans.
The page includes measurable elements (rates, comparisons) tied to outcomes.
2.2× The compute per dollar of AWS on-demand, 8× H100
Minimal jargon in the headline itself; the rest of the page uses some domain-specific terms.
Scored from the page as a whole; no single line was cited.
Next rungIf aiming higher, reduce abstractions elsewhere (e.g., avoid words like 'workspace' in the supporting copy).
No named customers or verifiable third-party data; only illustrative comparisons.
Illustrative comparisons using the AWS p5.48xlarge on-demand rate in us-east-1 ($55.04/h, October 2026) and Modal's July 2026 GPU list prices.
Next rungInclude a customer name with numbers or a verifiable usage metric.
There is an installation action presented, which serves as a next step.
Install curl -fsSL https://supercomputer.sh/install.sh | sh
Next rungFollow with a clearer CTA like 'Install now' or 'Get started now' tied to the product claim.
GPUs for agents, loved by humans.Shows the concrete product noun (GPUs) and target use (agents).
2.2× The compute per dollar of AWS on-demand, 8× H100Numerical proof tied to outcomes, meeting specificity criteria.
supercomputer — GPUs for agents, loved by humans GPUs for agents, loved by humans. Run inference, training, experiments, evals, and fine-tuning from your agent or terminal. Keep your workspace between runs. Install curl -fsSL https://supercomputer.sh/install.sh | sh Copy Installs the super CLI, signs you in, and teaches your coding agent to use it. More compute per dollar Typical rates, billed per second. 2.2× The compute per dollar of AWS on-demand, 8× H100 AWS supercomputer 20% Less than Modal on most GPUs Modal supercomputer $0 Compute when stopped Illustrative comparisons using the AWS p5.48xlarge on-demand rate in us-east-1 ($55.04/h, October 2026) and Modal's July 2026 GPU list prices. supercomputer is a compute market, so rates move; your live rate is shown when you provision. Workspace storage is metered separately.
Strongest: Clarity 95. Weakest: Proof 5. The 3 passes agreed within 4 points. The band 68–78 is where a re-score of this unchanged page would most likely land: on 42 dry re-scores (12 Oct 2026) the median move was 6 points and one in four moved more than 10. A grade only changes when a new score clears the boundary by 5.
No word in the headline itself was judged empty, yet jargon-free scores 65: the scorer found generic phrasing further down the page. Minimal jargon in the headline itself; the rest of the page uses some domain-specific terms.
Static SVG. No JavaScript, no tracking.
Scored from the live homepage
on 2026-10-14 by gpt-5-nano,
3 independent passes, median taken.
Page snapshot c5be1b98c506.
Rubric 7329c240 (current; what it says).
No person reviewed or edited this score. If the rubric and the score disagree, the rubric wins
and the page is re-scored — not hand-corrected.
Re-scored when the homepage text changes.