Banja Lab / Benchmarks / Test
The same task, run on 28 models. Compare the outputs side by side, or open any one in a popup to inspect it.
Top result: Yuki Tanabe (kimi-k3) (default reasoning) at 93.8% composite. Lowest: Noor El-Amin (claude-fable-5) at 0.0%. 28 models compared on this task.
# Client brief: Northshore Rapid Plumbing You are designing and building a single-page website for a real local service business. The brief below gives you the business facts, the goal, the audience, the sections the site must have, and the exact content to use. It gives you no visual direction on purpose. Every design decision - layout, type, colour, imagery, motion, tone - is yours to make. Read the whole brief, form a point of view on how this business should present itself, and build it. ## The business Northshore Rapid Plumbing is a licensed emergency plumbing and gasfitting business based in Chatswood, on Sydney's Lower North Shore, in New South Wales. It has run since 2009. It handles burst pipes, blocked drains, dead hot water systems, gas leaks, and the other problems that make a person stop what they are doing and look for a plumber right now. It runs a 24 hour, 7 day emergency line and aims to be on site within 60 minutes across the Lower North Shore. ## What the site is for The one job of this site is to turn a visitor into a phone call or a booking. Most people who reach it are on a phone, at home, with a problem happening in front of them - a pipe leaking, no hot water, a drain backing up - and they want to reach someone fast and trust that the person will turn up, do the work, and charge a fair, known price. Getting the visitor to call the number or send a booking enquiry is the point of every section. ## Who it is for Homeowners and renters on Sydney's Lower North Shore with an urgent or near-urgent plumbing problem, plus a smaller number of people booking non-urgent work (a dripping tap, a hot water replacement, a backflow test). Assume most arrive on a mobile phone. ## How people will use it This is a phone-first site. Its primary canvas is a mobile screen: assume a person holding a phone in one hand while water is on the floor. The phone number must be reach- able from anywhere on the page without hunting for it, and a person must be able to start a call or a booking in one tap. The design must read and work excellently at 360px wide, and it must remain complete and usable on a tablet and a desktop as well. These are requirements about how the site must function on a phone, not instructions about how it should look. One firm rule: any motion must respect a visitor's reduced-motion preference. When reduced motion is requested, the whole site must still be completely readable and every section reachable as a calm, static page, with no information conveyed only through movement. ## What the site must contain (all of these sections) - a hero that names the service and the suburb and carries a prominent call-now and book action, - the services offered, named, each with a short description and the indicative price where one is given, - why choose us / trust: the licence, the years in business, the guarantees, and the insurance, - reviews from named customers with star ratings, - the team, the named plumbers, - hours, location, and how to find and park, - a booking enquiry form, - contact details, with the phone number prominent throughout the page. ## Content to use Use the content below substantially as written, in Australian English. You may write your own section headings and short connective microcopy (button labels, captions, form labels), but the facts, figures, names, prices, and details are fixed. Do not invent extra services, drop any listed, or change the prices, the reviews, the team, the hours, the address, or the phone number. ### The pitch (use in or near the hero) Emergency plumber in Chatswood and across Sydney's Lower North Shore. Licensed, on site within 60 minutes, upfront fixed pricing agreed before any work starts. Open 24 hours, 7 days. Phone number, to appear prominently and repeatedly across the page: (02) 8095 4471. ### Services offered Present all eight. Each has a name, a short description, and an indicative starting price. - Burst and leaking pipes. Emergency isolation and repair of burst, split, and leaking pipes, indoors and out. From $180. - Blocked drains. CCTV camera inspection and high-pressure water jetting to clear blocked stormwater and sewer drains, including tree-root blockages. From $220. - Hot water systems. Repair or same-day replacement of gas, electric, and heat-pump hot water systems. Repairs from $150. New systems from $1,290 installed. - Blocked and leaking toilets. Unblock, re-seal, or replace a failed or overflowing toilet. From $160. - Tap and mixer repairs. Fix dripping taps and worn washers, or fit a new mixer. From $120. - Gas fitting. Locate and repair gas leaks, connect gas appliances, and issue compliance certificates. From $190. - Roof and gutter leaks. Locate and repair storm, downpipe, and gutter leaks. From $200. - Backflow testing. Annual backflow prevention testing and certification. From $165. There is no call-out fee when a repair is booked, and every job is quoted upfront before any work begins. ### Why choose us - Licensed NSW plumber and gasfitter, licence number 285194C. - Serving the Lower North Shore since 2009. - On site within 60 minutes for emergencies across Chatswood, Willoughby, Lane Cove, Artarmon, North Sydney, Naremburn, and Northbridge. - Upfront fixed pricing agreed before work starts. No hidden charges. - Fully insured, with $20 million public liability cover. - Lifetime workmanship guarantee on installations. ### Reviews Present these five named reviews with their star ratings. - Sarah Whitmore, Chatswood. 5 stars. "Burst pipe under the kitchen at 11pm and someone was here within the hour. Fixed it, showed me the price first, no surprises. Saved our floorboards." - David Nguyen, Lane Cove. 5 stars. "Hot water died on a Sunday. New unit installed the same day. Fair price, and they explained the options without pushing the dearest one." - Priya Menon, Artarmon. 5 stars. "A blocked drain that two other plumbers could not clear. They ran a camera, found tree roots, and jetted it out. Thorough and honest." - Tom Fletcher, Willoughby. 4 stars. "Quick to arrive and tidy work on a leaking toilet. It took one extra visit for a part, but they kept me posted the whole way." - Angela Brooks, North Sydney. 5 stars. "A gas leak scare, and they talked me through shutting it off on the phone before they arrived. Calm, licensed, and they sorted the compliance certificate too." ### The team Present these three named plumbers. - Mark Sullivan, founder and master plumber. Licensed plumber and gasfitter with 22 years in the trade. Started Northshore Rapid Plumbing in 2009 and still runs the emergency roster himself. - Jess Tanaka, licensed plumber and drainage specialist. Nine years in the trade. Leads the CCTV drain diagnostics and pipe relining work. - Ryan O'Dell, licensed plumber and gasfitter. Seven years in the trade. Handles hot water and gas appliance installs. ### Hours, location, and parking - Emergency line, open 24 hours, 7 days: (02) 8095 4471. - Office and bookings: Monday to Friday 7am to 5pm, Saturday 8am to 2pm. Sunday is emergency calls only. - Address: 12 Victoria Avenue, Chatswood NSW 2067. - Finding us and parking: a two-minute walk from Chatswood station. Metered street parking on Victoria Avenue, and free two-hour parking in the Westfield car park off Anderson Street. - Service area: Chatswood, Willoughby, Lane Cove, Artarmon, North Sydney, Naremburn, Northbridge, and the wider Lower North Shore. ### Booking enquiry form The site needs a booking enquiry form. It must collect: the visitor's name, phone number, email, preferred time, and which service they need (from the eight services above), plus a short field to describe the problem. Make clear near the form that for an emergency the fastest path is to call (02) 8095 4471 rather than wait on a form reply. ### Contact - Phone, 24 hours, 7 days: (02) 8095 4471. - Email: help@northshorerapid.com.au. - Address: 12 Victoria Avenue, Chatswood NSW 2067. ## Output contract (React, single page) Build one single-page site. Output exactly two files: - `App.tsx` - a default-exported React component. Define any subcomponents in this same file. This is the whole page. - `theme.css` - a Tailwind v4 `@theme` token block, your `font-family` choices from the shared font menu below, and any custom CSS you need. The only imports allowed in `App.tsx` are `react`, `framer-motion`, `lucide-react` (never the Sparkles icon), and, for 3D, `three`, `@react-three/fiber`, and `@react-three/drei`. No other packages, no network requests, and no external assets: all imagery must be inline SVG, CSS, drawn in code on a canvas, or rendered with WebGL. The fonts are bundled offline, so use only the shared menu: Archivo (including Archivo Expanded), Instrument Sans, Space Grotesk, Work Sans, IBM Plex Sans, IBM Plex Mono, JetBrains Mono, Fraunces, Instrument Serif, Playfair Display, Lora, Caveat. Make it responsive across mobile, tablet, and desktop. The page must also remain fully readable and usable if WebGL is unavailable: never leave a visitor with a blank screen or a missing section because a 3D canvas failed to start.
Anthropic: Every required section is present and correct: the "ON SITE IN 60." hero names the service and suburb with call-now and book actions, all eight services with descriptions and prices, the trust row (licence 285194C, since 2009, 60 min, $20m, lifetime guarantee), five named star-rated reviews, the three-person team, hours/location/parking, a full booking form with an emergency-call note, and the phone number repeated prominently (including the oversized footer treatment). The dark editorial system with a cyan accent, big Archivo Expanded display type, and a numbered mono-label grid is coherent,
Anthropic: Every required section is present and correct: hero names the service and suburb with call-now and book-online actions, all eight services with prices, why-choose-us including licence 285194C and $20M cover, five named reviews with correct star ratings (including Tom Fletcher's 4), three team members, hours/location/parking, a complete booking form with all specified fields plus the emergency-call note, and a contact footer with phone/email/address. The design is a confident editorial/brutalist system - cream, black, red and lime, with a striking mixed sans/serif hero and hard-shadow cards - t
Anthropic: Every required section is present and accurate: hero names the service and suburb with prominent call-now/book actions and a stats panel (licence 285194C, since 2009, $20M cover, 4.8 from 5), all eight services with correct prices, six trust points, five named reviews with star ratings, three team members, hours/address/parking, and a full booking form with the emergency call-out and prominent phone number repeated throughout. The navy-and-amber palette is coherent and trust-appropriate, typography and spacing are clean, and the sticky-feeling call CTA plus response-time panel show a clear pho
Anthropic: Every required section is present and correct: the hero names the service and Chatswood with prominent call and book actions, all eight services carry descriptions and starting prices, why-choose-us covers licence 285194C, years, insurance and guarantee, five named star reviews, the three plumbers, hours/location/parking with a map, a full booking form with the emergency-call callout, and the phone number repeats throughout including a persistent orange call bar. The navy-and-orange palette is coherent, urgent and appropriate for emergency service, with clean cards, a solid stat band and tidy
Anthropic: Every required section is present and correct: the hero names service and suburb with call/book actions, all eight services with their exact prices, the trust facts (licence 285194C, since 2009, $20M insurance, guarantee), five named reviews with star ratings, three team members, hours/location/parking, a full booking form with all fields and the emergency-call note, and the phone number repeated throughout including a bold red footer block. The execution is a confident Swiss/editorial layout with numbered sections, strong Archivo-style type hierarchy and a single red accent - coherent, clean,
Anthropic: Every required section is present and correct: an emergency hero naming service and suburb with call/book actions, all eight services with prices and "from" tags, the trust block (licence 285194C, 2009, $20M cover, 60-min promise), five named reviews with stars, the three named plumbers, hours/location/parking, a full booking form, and a phone number repeated throughout including a huge footer treatment. The visual language is a confident black/red/lime editorial-brutalist system with strong "Water won't WAIT" display type and a scrolling emergency ticker, which is distinctive and appropriate
Anthropic: Every required section is present and correct: hero names the service and Chatswood with call/book actions and phone, all eight services carry descriptions and starting prices, the trust block shows licence 285194C and $20M cover, five named star reviews, three named plumbers, hours/parking with a drawn map, a full booking form with the emergency-call note, and a phone-forward footer. The warm cream and peach palette with a serif display hero, hand-drawn van and badge illustrations reads as a coherent, friendly, editorial system that suits a local trade business. Craft is clean and consistent
Anthropic: Every required section is present and correct: hero naming service and suburb with call-now/book actions and a prominent phone number, all eight services with descriptions and starting prices, the trust bar (licence 285194C, since 2009, guarantees, $20m insurance), five named star-rated reviews, three named team members, hours/location/parking with a map, a complete booking form with the emergency-call callout, and a contact footer. The navy-and-orange system is coherent and professional, with clean cards, consistent iconography, a readable type hierarchy and a well-judged dark reviews section
Anthropic: Brief coverage is thorough and correct: hero names the service and Chatswood with call-now and book actions, all eight services appear with their prices, the why-choose-us trust block, five named star reviews, three team members, hours/location/parking, a full booking form with every required field plus the emergency-call note, and the phone number repeats throughout including the footer. The orange-and-navy palette is coherent and appropriately urgent for a plumbing trade, with clean card layouts and readable type. Craft is mostly solid but a couple of empty white pill/buttons (in the hero st
Anthropic: Every brief element is present and correct: dark hero naming the service and suburb with prominent call-now and book-online CTAs plus a repeated phone number, all eight services with descriptions and starting prices, the trust block with licence/insurance/guarantee, five named star reviews, three named team members, hours/location/parking, a complete booking form with the emergency call-first note, and a contact footer. The visual system is coherent and professional - navy base, teal accent, high-contrast orange CTAs, consistent card grids and generous spacing - and reads cleanly as a trustwor
Anthropic: Every required section is present and correct: hero names service and suburb with prominent call/book actions and repeated phone number, all eight services with descriptions and prices, six trust points with licence and $20M insurance, five named star reviews, three named team members, hours/location/parking with a drawn map, a full booking form, and a contact footer. The dark navy and orange palette is coherent and professional with clean card grids, sensible type hierarchy, and a persistent emergency phone CTA. Craft is consistent and finished though the map and some cards are fairly plain;
Anthropic: All required sections are present and correct: hero with the pitch, call/book actions and a nav phone number, a 60min/24-7/2009/$20M stat row, all eight priced services, the trust/licence block, five named reviews with stars, three team bios, hours/location/parking, a complete booking form with the emergency-call note, and a contact footer. The dark monochrome editorial treatment is coherent, consistently spaced and numbered, and looks modern, but it reads more like an agency portfolio than an urgent emergency plumber - CTAs are understated, body type is small and low-contrast, and the star ra
Anthropic: Every required section is present and correct: hero with call-now/book actions and the suburb, all eight services with their indicative prices, why-choose-us trust points, five named star reviews, three team members, hours/location/parking, a complete enquiry form with a service selector, and the phone number repeated prominently (top bar, a large red callout, hours, footer). The design is a coherent, restrained editorial layout - off-white ground, black type, single red accent, generous whitespace - which is tasteful and modern, though it reads quite flat and document-like with weak visual hi
Anthropic: Brief coverage is essentially complete: hero with service+suburb and Call/Book actions, all eight services with correct prices (including hot water's dual pricing), six trust points, five named star reviews, three team bios, hours/parking/service-area, a full booking form with the emergency-call callout, and a persistent top emergency bar plus footer keeping the phone number prominent throughout. The navy-and-orange palette is coherent and professional, with consistent cards, clear type hierarchy, and tidy alternating light/dark bands; spacing and detail look finished with no visible broken el
Anthropic: Every required section is present and correct: hero naming service and Chatswood with call-now and book actions, all eight services with descriptions and starting prices, why-choose-us with licence 285194C and $20M cover, five named reviews with stars, three named plumbers, hours/location/parking with an inline map, a complete booking form with the emergency-call caveat, and the phone number repeated throughout. The design is clean, coherent, and modern with a trustworthy blue palette, good spacing, a stats bar, and consistent card treatments, and craft is finished with no visible broken eleme
Anthropic: Brief coverage is complete and accurate: hero names the service and Chatswood with call/book actions and a 60min/24-7/$0 stat row, all eight services carry correct starting prices in a clean pricing list, and the trust block, five named star reviews, three-person team, hours/location/parking, full booking form, and a phone-prominent footer are all present. The visual language is coherent and modern - restrained white space, a disciplined red accent, and clear type hierarchy - though it reads as a safe, conventional service-site layout rather than something distinctive. Craft is solid with cons
Anthropic: All required sections are present and correct: hero naming service and suburb with call-now and book actions, the trust stat row (Since 2009, $20M, Lifetime), all eight named services with descriptions and prices, why-choose-us with licence/insurance/guarantee, five named star reviews, three named team members, hours/address/parking, a complete booking enquiry form with the emergency-call callout, and the phone number repeated in header, hero, CTA and footer. Visually it is clean, coherent and modern with a confident bold hero, consistent blue accent, tidy card grids and comfortable spacing; c
Anthropic: Brief coverage is comprehensive: hero with suburb and call/book CTAs, all eight services with correct prices, six trust points including licence 285194C and $20M cover, five named star-rated reviews, three team members, hours/location/parking, a complete booking form with the emergency-call note, and phone number repeated in header, hero, hours and footer. Visual design is clean, modern and coherent - a consistent blue-on-white card system with tidy spacing and clear typographic hierarchy - and craft is solid with uniform cards, icons and rendered star ratings, no obvious broken elements. It f
Anthropic: Brief fidelity is complete: hero with the pitch, 60-min target, all eight priced services, the six trust points, five named star reviews, three team members, hours/location/parking, a full booking form (name, phone, email, preferred time, service select, problem), and phone repeated in hero, hours, form and footer. The execution is a calm, editorial black-on-white serif treatment with a dotted-grid hero - coherent and consistent, but tonally mismatched to an urgent emergency-plumber whose one job is to drive a fast call; the "Call now" and "Book online" actions are tiny text links rather than
Anthropic: Brief coverage is complete and correct: editorial hero with call/book actions, all eight priced services, the six trust points (licence, since 2009, 60-minute arrival, fixed pricing, $20M liability, lifetime guarantee), five named star reviews, three team members, hours/location/parking, a full booking form with all required fields plus an emergency-call note, and the phone number repeated throughout. The warm cream and olive palette with the serif display type is coherent and reads editorial, but the heavy full-bleed black divider bars between sections look janky and blunt, and the call-now a
Anthropic: Content coverage is fairly complete - hero, all eight priced services, trust/licence block, five named star reviews, three team members, hours, map, booking form, and contact all appear - but the page renders as a mostly unstyled document: raw skip/nav links, flat left-aligned text, and no prominent, tappable call-now/book actions (they read as tiny inline links), which undercuts the brief's core phone-first goal. Visually it is incoherent, with three oversized flat colour rectangles (MS/JT/RO) dominating the team area and no consistent layout system, type hierarchy, or spacing. Craft looks br
Anthropic: Content fidelity is strong: all eight services with prices, the six trust points, five named star reviews, three team members, hours/parking, booking form, and the phone number all appear as specified. But the page renders essentially unstyled - default serif/sans browser fonts, tiny cramped single-column text, no cards, spacing system, colour, or layout hierarchy beyond blue links. Apart from a lone decorative pipe-and-droplet SVG at the top, this looks like a stylesheet that never applied; there is nothing premium, distinctive, or even finished about it, and it fails the phone-first, tap-to-
Anthropic: The content is essentially complete and correct: hero pitch with suburb and phone, all eight services with prices, why-choose-us with licence/insurance/guarantee, five named reviews with star rows, three team members, hours, location, parking, a booking form with the required fields, and full contact details. But the page renders as near-unstyled black-on-white body text with no layout system, no colour, no cards, and no hero treatment, so nothing reads as a designed site. Craft is badly broken - a stray grid artefact at the top, a collapsed map rendering as black blobs and white circles, and
Anthropic: Content fidelity is largely intact: the hero pitch, all eight named services with prices, the trust/licence block, five named reviews with star glyphs, the three team members, hours/location/parking, a booking form with the required fields, and the repeated phone number are all present. But visually the page is effectively unstyled - default browser serif headings, blue underlined links, plain bullet lists and raw form controls, with only a single stray orange circle in the hero standing in for a design. There is no layout system, colour, spacing, or hierarchy, so it reads as a broken render w
Anthropic: Most required content is present in source order - hero pitch with suburb and phone, the eight services with prices, why-choose-us licence/insurance facts, five named star reviews, three team members, hours/parking, a booking form and contact block - so factual coverage is moderate. But the visual layer has essentially failed: the page renders as tiny default-size unstyled text with no layout, no columns, no cards, no colour system, and no visible call-now buttons styled as such; star ratings show as empty outlines and a stray orange blob and faucet SVG float unmoored at the top. It looks like
Anthropic: The theme.css evidently failed to apply: the page renders as a wall of unstyled plain text with default link-blue coloring, no layout, no cards, no spacing, and no hierarchy - visually a broken render. Content fidelity is largely intact (all eight services with prices, five named reviews with star glyphs, the three team members, hours/parking, the booking form fields, and the repeated phone number are all present), which is the only thing keeping fidelity above the floor. But as a designed, phone-first conversion site it demonstrates essentially no visual design, craft, or impact - nothing a r
Anthropic: Most required content appears present (hero pitch, the eight services with prices, why-choose trust points, named reviews, the three team members, hours/location with a crude map, a booking form, and contact details), which earns partial fidelity. But the page is almost entirely unstyled: default browser typography, tiny left-aligned text, no layout system, no cards or spacing, and a decorative SVG pipe illustration that dominates while the rest reads like a raw document dump. Craft and impact are broken - the team section renders three enormous garish app-icon-style avatar blocks with generic