Design Evaluation

80 landing-page designs from 6 generation models across 10 style variations were each opened in a browser, captured at desktop and mobile viewports, and scored on a 10-criterion rubric (0–100 composite, ship / iterate / reject). Cross-model ranks were assigned per variation; model conclusions were synthesized from all judge write-ups.

Judge model: Composer 2.5 · Show more

Model conclusions

Model	Avg score	Best	Wins	Ship / Iter / Reject
Claude Opus 4.7	84	87 (v1)	2	8 / 2 / 0

A design-lead craft specialist: magazine editorial, heritage athletic club, expedition narrative, and Bauhaus systems—with near-perfect section completeness (9.90 avg) and the field's strongest visual identity scores.

Highest craft ceiling in the benchmark. Opus-4.7 leads on visual identity, completeness, and model average, making it ideal for premium launches where design-review pass rate matters. Pair with brand-energy guardrails (contemplative editorial undershoots Cadence's energetic brief) and mandatory mobile QA—v7 and v10 show the floor is still slop-prone without explicit constraints.

Consistent strengths

Highest model average (84.0) and visual identity (8.60) in the 80-entry dataset
Completeness leader at 9.90 — judges rarely found missing sections
Won variations 5 (heritage athletic) and 9 (expedition-themed) with thematic cohesion rare in AI output
80% ship rate (8/10) — only two iterate verdicts, both with clear remediation paths

Consistent weaknesses

Recycled '214k members' stat across batch — 'feels fabricated, hurts honesty lens'
Brand energy often contemplative vs energetic brief — brand-fit laggard at 7.70
Lowest mobile readiness (7.50); v10 scored 5 for vertical rotated pillar text
v7 purple/glass Y2K relapse (77) — 'edges toward AI-hyperpop slop'

Best moment: Variation 5 (87, ship, cross-model winner): Heritage athletic club with postcard testimonial treatment, triple CTA hero, and vintage identity that 'avoids every AI slop tell.'
Worst moment: Variation 10 (76, iterate): Maximalist high-energy page with sticky-note testimonials but 'vertical rotated text nearly unreadable at 390px — do not ship mobile as-is.'
Skill delta: N/A — single harness (claude-code baseline), no skill variant in dataset.

Model	Avg score	Best	Wins	Ship / Iter / Reject
Claude Opus 4.6+ skill variants	83	89 (v10)	6	13 / 7 / 0

The benchmark's most reliable conversion architect: full section funnels, populated stats, clear CTA pairs, and distinctive art-direction range across 10 style variations—with editorial cream, terracotta wellness, and brutalist peaks.

Still the default recommendation for landing pages when winning cross-model head-to-heads matters. Opus-4.6 wins more variations than any other model and ships 65% of entries. Skip frontend-design skill unless chasing a specific art-direction bet; baseline harness is the conversion-safe choice.

Consistent strengths

Won 6 of 10 cross-model variations (v2, v3, v6, v7, v8, v10) — most of any model
Highest ship count (13/20) and leads typography (8.60), polish (8.60), and brand-fit (8.25)
Baseline average 84.2 — strong conversion mechanics with product UI in winning entries
Distinctive range without the batch's worst slop failure (that belongs to Sonnet v4 at 70)

Consistent weaknesses

frontend-design skill underperforms baseline by 1.8 points on average
Purple-gradient relapses on baseline v3, v4, v9
Copy-only heroes common in editorial v1; ghost gold CTAs trade conversion for aesthetics
Skill variant on v5 (75) and v10 (81) pulled down otherwise winning formulas

Best moment: Variation 10 baseline (89, ship, cross-model winner): Cream editorial with Push Day workout card, category ticker, and complete conversion funnel — 'best-in-model.'
Worst moment: Variation 5 +frontend-design (75, iterate): Soft pastel blob with emoji feature icons — 'feels more meditation app than energetic fitness.'
Skill delta: Net-negative for Opus (-1.8 avg: baseline 84.2 vs skill 82.4). Skill increases visual bravery on v3 (+7) and v7 (+2) but devastates v5 (-9) and v10 (-8) with pastel blobs and softened conversion urgency. Default to baseline harness for ship decisions.

Model	Avg score	Best	Wins	Ship / Iter / Reject
Gpt 5.5	83	88 (v1)	1	6 / 4 / 0

A matured Codex designer: product-led dark fitness pages, luxury concierge gold craft, and brutalist performance systems—with fewer catastrophic slop failures and a doubled ship rate over GPT-5.4.

The recommended Codex model for landing pages. GPT-5.5 matches Opus-4.6 on average (83.0 vs 83.3), doubles its predecessor's ship rate, and won a cross-model variation. Pair with lifestyle photography on softer entries and enforce mobile nav retention—the main gap keeping it from Opus-4.7's craft ceiling.

Consistent strengths

Ship rate 60% (6/10) vs GPT-5.4's 30% — most consistent Codex harness
Won variation 4 outright with gold-on-black luxury concierge narrative
Strong completeness (9.10) and visual identity (8.40) — ties GPT-5.4 on copy (8.40)
Distinctive winners in brutalist v7 and night-race v10 without emoji heroes

Consistent weaknesses

Neon-on-black v1 still 'matches common AI dark-fitness templates'
Mobile nav collapse — center links hidden, only primary CTA remains at 390px
v8 analytics-SaaS aesthetic 'interchangeable with analytics SaaS' — lowest in-model at 79
Soft wellness tones on v6 and v9 undershoot energetic Cadence brief

Best moment: Variation 4 (88, ship, cross-model winner): Luxury concierge with gold craft, 91% proof stat tile, Personal Week Cycle 04 product card, and four named testimonials referencing specific product features.
Worst moment: Variation 8 (79, iterate): Data-forward page with 84.7 consistency index widget — 'competent but interchangeable with analytics SaaS; reads B2B more than lifestyle fitness.'
Skill delta: N/A — single harness (codex), no skill variant in dataset.

Model	Avg score	Best	Wins	Ship / Iter / Reject
Claude Sonnet 4.6+ skill variants	81	88 (v7)	0	11 / 9 / 0

High-ceiling editorial talent with a dangerous floor: magazine-grade peaks and the batch's worst purple-glass failure (70) coexist in the same model's 20 entries.

Best for editorial and lifestyle fitness brands when strictest anti-slop guardrails are enforced. Sonnet never won a cross-model variation outright despite two 88-point entries, revealing a gap between peak craft and competitive consistency. Always run visual-identity pre-check; never deploy frontend-design skill on gradient-prone briefs.

Consistent strengths

Editorial excellence — terracotta/sage palettes, magazine mastheads, warm organic wellness
v7 baseline and skill both at 88 — 'would pass a design-lead review today'
frontend-design skill net-positive (+1.7 avg) with standout retro product-data cards on v6 (+9)
Distinctive typographic moments judges called 'design-lead ready'

Consistent weaknesses

Zero sole cross-model variation wins across 20 entries — tied at v2 and v7 but never led outright
Widest identity failure mode — purple glassmorphism v4 (70–71), synthwave v6 (72)
Lowest polish average (8.00) and completeness (8.60) when skill misfires
Inconsistent brand-fit — pastel/playful and neon-dark both undershoot premium energy

Best moment: Variation 7 baseline (88, ship): Magazine masthead 'ISSUE NO. 6 — MARCH 2026' with subscribe band — 'standout editorial landing page.'
Worst moment: Variation 4 +frontend-design (70, iterate): Purple-pink glassmorphism — 'worst visual identity in batch; peak AI slop.'
Skill delta: Mildly net-positive (+1.7 avg: baseline 80.0 vs skill 81.7). Skill rescues v6 (+9) and v9 (+4) with bold retro craft but cannot fix slop-prone v4 (-1). Safer on baseline for variations with gradient risk.

Model	Avg score	Best	Wins	Ship / Iter / Reject
Claude Opus 4.8	81	86 (v3)	0	6 / 4 / 0

A wide-range craft generalist spanning neubrutalist systems, luxury editorial, telemetry dashboards, and lifestyle wellness—but ships less consistently than Opus-4.7, with six of ten clearing the bar and one purple-glass relapse dragging the average.

v3 neubrutalist and v7 telemetry heroes rival the benchmark's best, but 81.0 avg and no sole cross-model wins trail Opus-4.7. Ban purple-gradient glass — v4's identity score of 5 shows the relapse floor. Use with anti-slop guardrails; not a straight upgrade over 4.7 on consistency.

Consistent strengths

Peak craft on v3 neubrutalist (86, tied cross-model winner) and v7 telemetry hero (84) with product UI before signup
Best lifestyle brand-fit in the set on v6 (82) — warm approachable tone judges called closest to the Cadence brief
Luxury editorial v9 (84) with black-and-gold serif framing and impeccable typography
Tied for best mobile readiness in field (7.80 avg) alongside Opus-4.6

Consistent weaknesses

Zero sole cross-model variation wins — v3 tied at 86 with Opus-4.6 and 4.7 but never led outright
v4 purple-cyan gradient glassmorphism (71) — visual identity score of 5, lowest in the model set
Ship rate 60% (6/10) vs Opus-4.7's 80% — v2 missing testimonial band, v8 copy typo, v10 pastel template feel
Content/copy laggard at 7.90 — synthwave v5 and sparse luxury v9 heroes lack product proof above the fold

Best moment: Variation 3 (86, ship, tied cross-model winner): Cohesive neubrutalist system with hero dashboard proving the product immediately, per-feature CTAs, and complete conversion funnel.
Worst moment: Variation 4 (71, iterate): Purple-cyan gradient glassmorphism — 'technically complete but visually indistinguishable from countless AI-generated gradient SaaS pages; fails the anti-slop and craft bar.'
Skill delta: N/A — single harness (claude-code baseline), no skill variant in dataset.

Model	Avg score	Best	Wins	Ship / Iter / Reject
Gpt 5.4	79	88 (v1)	1	3 / 7 / 0

A volatile product-data storyteller: peaks rival the best Claude output on credible stats and asymmetric trust signals, but relapses into pastel card mosaics, neon cyber gradients, and low-energy editorial pages.

High-ceiling, low-floor Codex baseline. Judges valued its data-rich heroes and credible copy when anti-slop guardrails held, but only 3 of 10 cleared ship. Use with explicit bans on neon gradients and pastel card grids; GPT-5.5 largely supersedes it on consistency without sacrificing product storytelling.

Consistent strengths

Best average content/copy score among GPT models (8.40) — specific stats with named testimonials
Strong product-data storytelling via dashboard widgets, challenge cards, and 87% retention tiles
Honesty gate passed consistently — no zeros or lorem across all 10 variations
Tied variation 1 winner with product-led dark navy hero at 88

Consistent weaknesses

Lowest visual identity (7.60) and brand-fit (7.20) in the full 80-entry field
Thin footers — 'single line with email only' appears repeatedly
Only 30% ship rate (3/10) — highest variance in the Codex line (71–88)
Neon cyber and pastel card-grid relapses on v6, v9, v10

Best moment: Variation 1 (88, ship, cross-model winner): Dark navy hero with teal dashboard card, populated 120k+ workout stats, and asymmetric 87% retention testimonial tile.
Worst moment: Variation 9 (71, iterate): Neon cyan-green cyber aesthetic — 'would get rejected in a design review for feeling generated.'
Skill delta: N/A — single harness (codex), no skill variant in dataset.

Individual design reviews

Show:

Findings by variation

Variation 1— winner: Gpt 5.4 (88)

8 designs compared

Gpt 5.4

codex · variation 1

ship

Best-in-model: product-led hero with credible data storytelling and asymmetric social proof.

Strengths

Hero dashboard widget shows weekly consistency bars and active challenge
87% retention stat in testimonial grid is a smart asymmetric trust signal
All hero stats populated — 120k+ workouts, not zeros
Dark teal palette avoids generic AI purple-gradient fitness cliché

Weaknesses

Teal-on-navy is competent but safe — limited brand memorability beyond craft
Footer is a single line with email only — no secondary nav links

Stakeholder-ready. The page demonstrates the product in the hero before asking for signup, then earns trust with asymmetric social proof. The 87% retention tile beside testimonial quotes is senior-level landing page thinking.

View design →Model page →Desktop screenshot →

▸7 other designs in this variation

#2 Gpt 5.5UX88

Product-led dark landing page with credible stats, live-plan hero widget, and asymmetric social proof.

#3 Claude Opus 4.7UX87

A magazine-quality landing page that would pass a design review — distinctive editorial voice with every required section executed with care.

#4 Claude Opus 4.6UX85

Luxury editorial landing page with strong brand voice and complete section delivery.

#5 Claude Opus 4.6UX83

Textbook high-converting landing page — safe but expertly built.

#6 Claude Opus 4.8UX82

A shippable dark athletic landing page with strong hierarchy, complete sections, and credible product visualization.

#7 Claude Sonnet 4.6UX81

Complete dark-neon conversion page — functional but not distinctive enough to lead.

#8 Claude Sonnet 4.6UX81

Best hero photography undermined by contrast failures below fold.

Variation 2— winner: Claude Opus 4.6 (86)

8 designs compared

Claude Opus 4.6

claude-code · variation 2

ship

Editorial wellness landing page with distinctive terracotta palette and product progress widget.

Strengths

Italic terracotta 'refined' in serif headline is a memorable brand moment
THIS WEEK'S PROGRESS card with 72% bar demonstrates tracking value
Scrolling category ticker adds motion without ticker-slop cliché
Journal nav link signals editorial brand depth

Weaknesses

Terracotta-on-cream is distinctive but adjacent to other warm editorial entries
Testimonials are solid but lack asymmetric layout surprise

Stakeholder-ready for a wellness-forward Cadence positioning. The progress card earns trust before the features section, which is exactly how a design lead wants a fitness SaaS hero to work.

View design →Model page →Desktop screenshot →

▸7 other designs in this variation

#2 Claude Sonnet 4.6UX86

Warm organic wellness page — complete, credible, and distinctly non-AI.

#3 Claude Sonnet 4.6UX86

Skill-enhanced warm organic — same A-tier completeness as base v2.

#4 Claude Opus 4.7UX86

High-energy brutalist landing that nails Cadence's energetic positioning with strong CTAs and complete structure — dense on mobile but shippable.

#5 Gpt 5.5UX83

Editorial magazine landing page with strong weekly-plan demo and credible member notes.

#6 Claude Opus 4.8UX83

The most refined editorial direction in the set — premium and distinctive — held back only by a missing testimonial band.

#7 Claude Opus 4.6UX79

Cyberpunk fitness page with strong visual effects but bro-marketing copy and sparse hero.

#8 Gpt 5.4UX73

Elegant editorial page with strong copy stats but low energy for the Cadence brief.

Variation 3— winner: Claude Opus 4.6 (86)

8 designs compared

Claude Opus 4.6(frontend-design)

claude-code · variation 3

ship

Organic wellness landing page with distinctive terracotta palette and complete conversion funnel.

Strengths

Script 'Rhythm' in headline is a memorable brand moment
ROOTED IN WELLNESS sage badge sets tone before headline
Holistic tracking copy (sleep, mood, recovery) differentiates from rep-count competitors
Full three-column footer with Product/Company/Support architecture

Weaknesses

Abstract circle graphic is decorative — no product UI preview
Organic wellness palette may feel soft for 'energetic' brief interpretation

Shippable for a wellness sub-brand. The frontend-design skill clearly pushed visual originality here without sacrificing section completeness.

View design →Model page →Desktop screenshot →

▸7 other designs in this variation

#2 Claude Opus 4.7UX86

Refined calm-fitness landing with excellent mobile behavior and micro-product UI — would ship for a wellness sub-brand, slightly soft for core Cadence energy.

#3 Claude Opus 4.8UX86

Top pick — a fully realized conceptual system that is distinctive, complete, and conversion-aware.

#4 Claude Sonnet 4.6UX82

Pastel social-club landing — complete, approachable, mobile-strong.

#5 Claude Sonnet 4.6UX82

Structured dark-tech landing with product intelligence cards — shippable with caveats.

#6 Gpt 5.5UX82

Distinctive track-club landing page with relay leaderboard hero and credible challenge stats.

#7 Claude Opus 4.6UX79

Complete dark SaaS landing page held back by purple-gradient anti-slop penalties.

#8 Gpt 5.4UX79

Energetic drive-forward page with strong product proof in the hero — needs footer and mobile polish.

Variation 4— winner: Gpt 5.5 (88)

8 designs compared

Gpt 5.5

codex · variation 4

ship

Luxury concierge landing page with gold craft, 91% proof stat, and complete premium narrative.

Strengths

Gold-on-black concierge identity avoids AI gradient slop
91% stat tile anchors social proof section
Personal Week Cycle 04 product card in hero
Four named testimonials with specific product references

Weaknesses

Ghost gold CTA is beautiful but weaker conversion than solid fills
Premium calm tone may feel slow for high-energy campaigns

Clear winner within gpt-5.5. I'd ship this for a premium tier launch — the dark-to-cream scroll narrative is senior-level craft.

View design →Model page →Desktop screenshot →

▸7 other designs in this variation

#2 Claude Opus 4.6UX83

Brutalist fitness page with unmistakable visual identity and complete conversion path.

#3 Gpt 5.4UX82

Calm-precision landing page that demonstrates the product clearly and earns trust with named testimonials.

#4 Claude Opus 4.7UX82

Technically polished dev-OS landing that clears the ship bar on craft and completeness — narrow audience fit keeps it from top rank.

#5 Claude Opus 4.6UX81

Competent SaaS landing page with strong product preview but generic gradient aesthetic.

#6 Claude Sonnet 4.6UX71

Functionally complete purple-gradient page — fails visual identity bar.

#7 Claude Opus 4.8UX71

Technically complete but visually indistinguishable from countless AI-generated gradient SaaS pages.

#8 Claude Sonnet 4.6UX70

frontend-design skill doubled down on purple glass — worst visual identity in batch.

Variation 5— winner: Claude Opus 4.7 (87)

8 designs compared

Claude Opus 4.7

claude-code · variation 5

ship

Heritage athletic club landing with top-tier craft and conversion paths — a design lead would ship this for a premium fitness launch.

Strengths

Postcard testimonial treatment is memorable and on-brand
Triple CTA hero gives clear conversion options
Vintage identity avoids every AI slop tell

Weaknesses

Repeated 214k member stat across batch
Yellow 'BROWSE THE CATALOG' button contrast could improve

Minor button contrast fix on secondary CTA. Strong candidate for A/B against v1 editorial direction.

View design →Model page →Desktop screenshot →

▸7 other designs in this variation

#2 Claude Sonnet 4.6UX86

Biophilic nature design with product UI preview — shippable wellness positioning.

#3 Gpt 5.4UX86

Luxury-tier landing page with the strongest product-data storytelling after v1.

#4 Claude Sonnet 4.6UX85

Earthy split-screen with credible stats and complete scroll architecture.

#5 Claude Opus 4.6UX84

Energetic split-layout page with live workout checklist and credible social proof.

#6 Gpt 5.5UX80

Command-center fitness page with strong protocol widget and populated telemetry stats.

#7 Claude Opus 4.8UX76

A fun, fully themed synthwave page that impresses as a concept but feels too gimmicky for a premium fitness launch.

#8 Claude Opus 4.6UX75

Gentle wellness page with complete sections but generic pastel aesthetic and soft energy.

Variation 6— winner: Claude Opus 4.6 (88)

8 designs compared

Claude Opus 4.6

claude-code · variation 6

ship

Product-led lifestyle landing page with activity rings and complete member proof.

Strengths

Weekly progress card (5 of 6 sessions) and three-ring activity widget in hero
Wearable integration bar (Apple Watch, Garmin, WHOOP, Strava, Fitbit)
Six feature cards covering programming through device sync
Physical therapist testimonial adds unexpected credibility

Weaknesses

Pill nav bar pattern is trendy but slightly overused in SaaS
Green-on-cream is safe — limited brand memorability beyond craft

Solid A-tier work I'd ship today. Every section earns its scroll position. The activity rings immediately communicate what Cadence tracks, which is senior-level landing page thinking.

View design →Model page →Desktop screenshot →

▸7 other designs in this variation

#2 Claude Opus 4.7UX86

Bauhaus-system landing with strong conceptual framework and production polish — shippable with confident design-lead approval.

#3 Claude Opus 4.6UX83

Retro groovy landing page with strong personality and complete member proof.

#4 Claude Opus 4.8UX82

A warm, lifestyle-forward page that best embodies the Cadence brand brief and ships with complete structure.

#5 Claude Sonnet 4.6UX81

Bold 80s retro with live product data cards — iterate on scroll completeness.

#6 Gpt 5.5UX80

Warm friendly landing page with habit-focused copy and credible community stat.

#7 Gpt 5.4UX73

Colorful and complete but the pastel card-grid hero triggers anti-slop concerns.

#8 Claude Sonnet 4.6UX72

Synthwave spectacle with decent scroll depth — iterate on copy and conversion clarity.

Variation 7— winner: Claude Opus 4.6 (88)

8 designs compared

Claude Opus 4.6(frontend-design)

claude-code · variation 7

ship

Swiss brutalist fitness page with unmistakable visual identity and strong conversion path.

Strengths

Red 'REP' word in stacked headline is instantly memorable
Numbered 01/02/03 feature structure with coach-certified copy
Marcus Chen software-engineer testimonial in all-caps adds personality
Red left border creates editorial frame without clutter

Weaknesses

All-caps density in lower sections may fatigue on long scroll
No product UI preview — entirely copy-driven hero

This would pass a design lead review for a performance-focused sub-brand. The Swiss aesthetic is genuinely differentiated from the lime-on-black and purple-gradient entries in this model set.

View design →Model page →Desktop screenshot →

▸7 other designs in this variation

#2 Claude Sonnet 4.6UX88

Standout editorial landing page — would pass a design-lead review today.

#3 Claude Sonnet 4.6UX88

Luxury monochrome take on the winning editorial structure — equally shippable.

#4 Claude Opus 4.6UX86

High-performance dark athletic page with complete feature story and athlete credibility.

#5 Gpt 5.5UX84

Brutalist fitness page with unmistakable visual identity and hard progress proof.

#6 Claude Opus 4.8UX84

Near-top execution — product proof in the hero is the strongest conversion argument in the model set after v3.

#7 Gpt 5.4UX79

Memorable brutalist grid with excellent hierarchy — premium brief fit is the gap.

#8 Claude Opus 4.7UX77

Competent glossy landing that hits section checklist but triggers anti-slop concerns and weakens on mobile — iterate before ship.

Variation 8— winner: Claude Opus 4.6 (86)

8 designs compared

Claude Opus 4.6

claude-code · variation 8

ship

Mindful wellness landing page with warm teal palette and credible progress widgets.

Strengths

23-day streak and 78% monthly progress cards demonstrate tracking without dashboard overload
Philosophy section ('fitness should feel like freedom') earns the mindful positioning
Adaptive Growth Plans feature with 🌱 icon fits brand without emoji-hero slop
Three testimonials with star ratings and role labels

Weaknesses

Teal-on-cream wellness palette is polished but not radically unique
Progress cards overlap visually — could feel busy on smaller screens

This passes the design-lead gate for a wellness-forward Cadence variant. The intention-focused copy and progress widgets align tone with action.

View design →Model page →Desktop screenshot →

▸7 other designs in this variation

#2 Claude Opus 4.7UX86

Exceptionally crafted zen-fitness landing with best-in-class mobile behavior — shippable as a premium variant though energy is muted vs. brief.

#3 Claude Opus 4.6UX84

Art-deco luxury landing page with strong brand atmosphere and complete sections.

#4 Claude Sonnet 4.6UX82

Brutalist typographic hero with credible stats — distinctive and nearly complete.

#5 Claude Opus 4.8UX81

Visually bold and complete, but copy typos and mobile overlap keep it just below the ship threshold.

#6 Claude Sonnet 4.6UX80

Dark gold luxury template — complete but not distinctive enough.

#7 Gpt 5.5UX79

Data-forward landing page with strong stats but generic analytics-SaaS visual language.

#8 Gpt 5.4UX76

Well-crafted wellness landing page that misses the energetic Cadence positioning.

Variation 9— winner: Claude Opus 4.7 (87)

8 designs compared

Claude Opus 4.7

claude-code · variation 9

ship

Expedition-themed landing with narrative cohesion and production craft — second-ranked in the set and fully shippable.

Strengths

Thematic consistency from hero seal to trail chart is rare and effective
Community activity log adds live-product credibility
Gold-on-green palette feels premium, not generic

Weaknesses

Dark theme may feel heavy for broad fitness audience
Some expedition jargon ('Camp 5') adds cognitive load

A/B test against v1 for conversion. Consider lightening one mid-page section for scroll rhythm.

View design →Model page →Desktop screenshot →

▸7 other designs in this variation

#2 Claude Opus 4.8UX84

A luxury editorial direction with exceptional craft — ships for premium positioning despite a sparse above-fold hero.

#3 Gpt 5.5UX81

Coastal lifestyle landing page with balanced structure and credible community proof.

#4 Claude Opus 4.6UX80

Strong product preview undermined by purple-gradient AI aesthetic.

#5 Claude Opus 4.6UX80

High-energy neo-brutalist page with strong CTAs but intense palette and ticker gimmick.

#6 Claude Sonnet 4.6UX76

Structurally complete pastel page marred by emoji-heavy AI tells.

#7 Claude Sonnet 4.6UX72

Playful colorful hero with emoji metrics — iterate before shipping.

#8 Gpt 5.4UX71

Data-rich cyber landing page held back by neon gradient anti-slop signals.

Variation 10— winner: Claude Opus 4.6 (89)

8 designs compared

Claude Opus 4.6

claude-code · variation 10

ship

Best-in-model: editorial cream layout with live product mockup and complete conversion funnel.

Strengths

Push Day workout card with five exercises demonstrates the product before signup
22/28 monthly progress bar with On Track badge is credible product UI
Category ticker (CONDITIONING · MOBILITY · RUNNING) adds motion without gimmicks
All hero stats populated — 250K+, 4.9★, 12M workouts

Weaknesses

Cream-and-green palette is refined but not radically distinctive within fitness SaaS
Footer is functional but lighter than the hero craft level

This is stakeholder-ready. The page shows the actual training experience in the hero before asking for signup, then earns trust with testimonials. The mixed-type headline and session card elevate it above copy-only competitors in this model set.

View design →Model page →Desktop screenshot →