# Capyboard

> A curated, inspectable gallery of AI model responses to one HTML, CSS, and JavaScript coding challenge: animate a capybara on a hoverboard. Not a model leaderboard.

Capyboard preserves each collected AI model attempt alongside its exact prompt, recorded generation conditions, token and timing metrics, SHA-256 artifact digests, and qualitative curator observations.

## Key Pages & Machine-Readable Endpoints

- [Gallery](https://capy.ai.id/): Browse and filter all published model attempts.
- [The Challenge & Protocol](https://capy.ai.id/challenge/): Exact prompt text, SHA-256 checksums, collection protocol, and restricted preview policy.
- [Side-by-Side Comparison](https://capy.ai.id/compare/): Compare any two published runs across equal 4:3 viewports and recorded metrics.
- [Published Runs Catalogue (JSON)](https://capy.ai.id/assets/runs.json): Machine-readable JSON catalogue of all published runs, measurements, curator observations, and evidence URLs.
- [Sitemap (XML)](https://capy.ai.id/sitemap.xml): Canonical URL index.

## Challenge Prompts

- [Capybara on a hoverboard (capybara-hoverboard-v1)](https://capy.ai.id/evidence/challenges/capybara-hoverboard-v1/prompt.txt): SHA-256 `e6b955ce7c23afabd2adf2115467e7b6c213ac2cab2a7596886ca71ad827fe1c`

## Published Runs

- [gpt-6-astra](https://capy.ai.id/runs/pi-01a13993-0d7d-71b8-9ae0-e251d6760d36-capybara-hoverboard-v1/) (`pi-01a13993-0d7d-71b8-9ae0-e251d6760d36-capybara-hoverboard-v1`, openai-codex, one-shot, Pi (high effort); Model time: 204.789 s, Input tokens: 1,262, Cached tokens: 0, Output tokens: 7,534, Total tokens: 8,796, Cost (USD): Unknown) — Strong one-shot execution: a single model call with no tool use produced a complete, polished experience.
  Evidence: [run.json](https://capy.ai.id/evidence/runs/pi-01a13993-0d7d-71b8-9ae0-e251d6760d36-capybara-hoverboard-v1/run.json), [curation.json](https://capy.ai.id/evidence/curation/pi-01a13993-0d7d-71b8-9ae0-e251d6760d36-capybara-hoverboard-v1.json), [transcript.json](https://capy.ai.id/evidence/runs/pi-01a13993-0d7d-71b8-9ae0-e251d6760d36-capybara-hoverboard-v1/transcript.json), [response.txt](https://capy.ai.id/evidence/runs/pi-01a13993-0d7d-71b8-9ae0-e251d6760d36-capybara-hoverboard-v1/response.txt), [HTML source](https://capy.ai.id/evidence/runs/pi-01a13993-0d7d-71b8-9ae0-e251d6760d36-capybara-hoverboard-v1/output.html.txt)
- [deepseek-flash](https://capy.ai.id/runs/pi-01a13428-91a4-73e0-bb65-7ae074cb59a7-capybara-hoverboard-v1-cost/) (`pi-01a13428-91a4-73e0-bb65-7ae074cb59a7-capybara-hoverboard-v1-cost`, deepseek, tool-assisted, Pi (high effort); Model time: 96.369 s, Input tokens: 9,949, Cached tokens: 66,176, Output tokens: 22,483, Total tokens: 98,608, Cost (USD): $0.03) — The run used five model calls and 96,369 ms of model time.
  Evidence: [run.json](https://capy.ai.id/evidence/runs/pi-01a13428-91a4-73e0-bb65-7ae074cb59a7-capybara-hoverboard-v1-cost/run.json), [curation.json](https://capy.ai.id/evidence/curation/pi-01a13428-91a4-73e0-bb65-7ae074cb59a7-capybara-hoverboard-v1-cost.json), [transcript.json](https://capy.ai.id/evidence/runs/pi-01a13428-91a4-73e0-bb65-7ae074cb59a7-capybara-hoverboard-v1-cost/transcript.json), [response.txt](https://capy.ai.id/evidence/runs/pi-01a13428-91a4-73e0-bb65-7ae074cb59a7-capybara-hoverboard-v1-cost/response.txt), [HTML source](https://capy.ai.id/evidence/runs/pi-01a13428-91a4-73e0-bb65-7ae074cb59a7-capybara-hoverboard-v1-cost/output.html.txt)
- [deepseek-v4-flash](https://capy.ai.id/runs/dsh-aacd7d25-7533-495a-8224-2d2881c24a42-capybara-hoverboard-v1-elapsed/) (`dsh-aacd7d25-7533-495a-8224-2d2881c24a42-capybara-hoverboard-v1-elapsed`, deepseek-official, tool-assisted, DeepSeek Harness (high effort); Elapsed time: 320 s, Input tokens: 22,647, Cached tokens: 1,078,144, Output tokens: 55,617, Total tokens: 1,156,408, Cost (USD): $0.04) — The run demonstrates strong iterative tool use: after image inspection failed, the model compensated with screenshots, pixel analysis, and layout measurements, then refined the scene and perspective grid. Its visual QA was indirect, so human review remains important before enabling the preview.
  Evidence: [run.json](https://capy.ai.id/evidence/runs/dsh-aacd7d25-7533-495a-8224-2d2881c24a42-capybara-hoverboard-v1-elapsed/run.json), [curation.json](https://capy.ai.id/evidence/curation/dsh-aacd7d25-7533-495a-8224-2d2881c24a42-capybara-hoverboard-v1-elapsed.json), [transcript.json](https://capy.ai.id/evidence/runs/dsh-aacd7d25-7533-495a-8224-2d2881c24a42-capybara-hoverboard-v1-elapsed/transcript.json), [response.txt](https://capy.ai.id/evidence/runs/dsh-aacd7d25-7533-495a-8224-2d2881c24a42-capybara-hoverboard-v1-elapsed/response.txt), [HTML source](https://capy.ai.id/evidence/runs/dsh-aacd7d25-7533-495a-8224-2d2881c24a42-capybara-hoverboard-v1-elapsed/output.html.txt)
- [deepseek-v4-flash](https://capy.ai.id/runs/dsh-adef23f3-d235-4306-93d7-9afaf9df31cf-capybara-hoverboard-v1-elapsed/) (`dsh-adef23f3-d235-4306-93d7-9afaf9df31cf-capybara-hoverboard-v1-elapsed`, deepseek-official, tool-assisted, DeepSeek Harness (high effort); Elapsed time: 142 s, Input tokens: 7,123, Cached tokens: 103,424, Output tokens: 33,648, Total tokens: 144,195, Cost (USD): $0.04) — The self-contained scene renders a capybara on a glowing cyan hoverboard against a sunset mountain landscape, with animated stars, clouds, thrusters, particles, and a red scarf.
  Evidence: [run.json](https://capy.ai.id/evidence/runs/dsh-adef23f3-d235-4306-93d7-9afaf9df31cf-capybara-hoverboard-v1-elapsed/run.json), [curation.json](https://capy.ai.id/evidence/curation/dsh-adef23f3-d235-4306-93d7-9afaf9df31cf-capybara-hoverboard-v1-elapsed.json), [transcript.json](https://capy.ai.id/evidence/runs/dsh-adef23f3-d235-4306-93d7-9afaf9df31cf-capybara-hoverboard-v1-elapsed/transcript.json), [response.txt](https://capy.ai.id/evidence/runs/dsh-adef23f3-d235-4306-93d7-9afaf9df31cf-capybara-hoverboard-v1-elapsed/response.txt), [HTML source](https://capy.ai.id/evidence/runs/dsh-adef23f3-d235-4306-93d7-9afaf9df31cf-capybara-hoverboard-v1-elapsed/output.html.txt)
- [muse-spark-1.3-contributor](https://capy.ai.id/runs/pi-01a1325c-8b2b-71f3-94c2-e752d9cf55b7-capybara-hoverboard-v1-cost-reported-cost/) (`pi-01a1325c-8b2b-71f3-94c2-e752d9cf55b7-capybara-hoverboard-v1-cost-reported-cost`, meta, tool-assisted, Pi (xhigh effort); Model time: 203.472 s, Input tokens: 16,436, Cached tokens: 5,474, Output tokens: 25,148, Total tokens: 47,058, Cost (USD): $0.006684) — The response is a complete self-contained HTML document with inline CSS and JavaScript; no external asset tags, imports, or network calls were found during source inspection.
  Evidence: [run.json](https://capy.ai.id/evidence/runs/pi-01a1325c-8b2b-71f3-94c2-e752d9cf55b7-capybara-hoverboard-v1-cost-reported-cost/run.json), [curation.json](https://capy.ai.id/evidence/curation/pi-01a1325c-8b2b-71f3-94c2-e752d9cf55b7-capybara-hoverboard-v1-cost-reported-cost.json), [transcript.json](https://capy.ai.id/evidence/runs/pi-01a1325c-8b2b-71f3-94c2-e752d9cf55b7-capybara-hoverboard-v1-cost-reported-cost/transcript.json), [response.txt](https://capy.ai.id/evidence/runs/pi-01a1325c-8b2b-71f3-94c2-e752d9cf55b7-capybara-hoverboard-v1-cost-reported-cost/response.txt), [HTML source](https://capy.ai.id/evidence/runs/pi-01a1325c-8b2b-71f3-94c2-e752d9cf55b7-capybara-hoverboard-v1-cost-reported-cost/output.html.txt)
- [gpt-oss-120b-medium](https://capy.ai.id/runs/agy-5ef17ce4-44c5-4d41-8e84-ebf25ed93d3c-capybara-hoverboard-v1/) (`agy-5ef17ce4-44c5-4d41-8e84-ebf25ed93d3c-capybara-hoverboard-v1`, google-vertex, tool-assisted, Antigravity CLI (medium effort); Model time: 10.721 s, Input tokens: 9,428, Cached tokens: 2,112, Output tokens: 2,113, Total tokens: 13,653, Cost (USD): Unknown) — A compact self-contained HTML artifact uses inline SVG and CSS keyframes to animate a capybara riding a hoverboard.
  Evidence: [run.json](https://capy.ai.id/evidence/runs/agy-5ef17ce4-44c5-4d41-8e84-ebf25ed93d3c-capybara-hoverboard-v1/run.json), [curation.json](https://capy.ai.id/evidence/curation/agy-5ef17ce4-44c5-4d41-8e84-ebf25ed93d3c-capybara-hoverboard-v1.json), [transcript.json](https://capy.ai.id/evidence/runs/agy-5ef17ce4-44c5-4d41-8e84-ebf25ed93d3c-capybara-hoverboard-v1/transcript.json), [response.txt](https://capy.ai.id/evidence/runs/agy-5ef17ce4-44c5-4d41-8e84-ebf25ed93d3c-capybara-hoverboard-v1/response.txt), [HTML source](https://capy.ai.id/evidence/runs/agy-5ef17ce4-44c5-4d41-8e84-ebf25ed93d3c-capybara-hoverboard-v1/output.html.txt)
- [gemini-3.8-flash-high](https://capy.ai.id/runs/agy-9716c6ed-b142-495a-9a4d-234569ea329e-capybara-hoverboard-v1/) (`agy-9716c6ed-b142-495a-9a4d-234569ea329e-capybara-hoverboard-v1`, google-vertex, tool-assisted, Antigravity CLI (high effort); Time: Unknown, Input tokens: 228,128, Cached tokens: 2,811,068, Output tokens: 124,002, Total tokens: 3,163,198, Cost (USD): Unknown) — The agent began with an SVG concept and then pivoted to a detailed Canvas implementation.
  Evidence: [run.json](https://capy.ai.id/evidence/runs/agy-9716c6ed-b142-495a-9a4d-234569ea329e-capybara-hoverboard-v1/run.json), [curation.json](https://capy.ai.id/evidence/curation/agy-9716c6ed-b142-495a-9a4d-234569ea329e-capybara-hoverboard-v1.json), [transcript.json](https://capy.ai.id/evidence/runs/agy-9716c6ed-b142-495a-9a4d-234569ea329e-capybara-hoverboard-v1/transcript.json), [response.txt](https://capy.ai.id/evidence/runs/agy-9716c6ed-b142-495a-9a4d-234569ea329e-capybara-hoverboard-v1/response.txt), [HTML source](https://capy.ai.id/evidence/runs/agy-9716c6ed-b142-495a-9a4d-234569ea329e-capybara-hoverboard-v1/output.html.txt)
- [gemini-3.8-flash](https://capy.ai.id/runs/pi-01a128ec-d942-7211-ac9e-19e0b5634a43-reported-cost/) (`pi-01a128ec-d942-7211-ac9e-19e0b5634a43-reported-cost`, google-vertex, tool-assisted, Pi (high effort); Model time: 980.485 s, Input tokens: 341,369, Cached tokens: 4,970,986, Output tokens: 152,332, Total tokens: 5,464,687, Cost (USD): $1.200096) — This high-effort Pi run used Gemini 3.8 Flash through Google Vertex and produced a 42.9 KB self-contained HTML response in 16 minutes 20 seconds, with 5.46M recorded total tokens.
  Evidence: [run.json](https://capy.ai.id/evidence/runs/pi-01a128ec-d942-7211-ac9e-19e0b5634a43-reported-cost/run.json), [curation.json](https://capy.ai.id/evidence/curation/pi-01a128ec-d942-7211-ac9e-19e0b5634a43-reported-cost.json), [transcript.json](https://capy.ai.id/evidence/runs/pi-01a128ec-d942-7211-ac9e-19e0b5634a43-reported-cost/transcript.json), [response.txt](https://capy.ai.id/evidence/runs/pi-01a128ec-d942-7211-ac9e-19e0b5634a43-reported-cost/response.txt), [HTML source](https://capy.ai.id/evidence/runs/pi-01a128ec-d942-7211-ac9e-19e0b5634a43-reported-cost/output.html.txt)
- [gemini-3.8-flash](https://capy.ai.id/runs/pi-01a12625-8eb6-72a8-8fe0-31810bac58d6-reported-cost/) (`pi-01a12625-8eb6-72a8-8fe0-31810bac58d6-reported-cost`, google-vertex, tool-assisted, Pi (medium effort); Model time: 575.1 s, Input tokens: 327,810, Cached tokens: 1,742,054, Output tokens: 86,216, Total tokens: 2,156,080, Cost (USD): $0.699822) — Self-contained HTML with inline CSS and JavaScript; no external assets or network requests were found.
  Evidence: [run.json](https://capy.ai.id/evidence/runs/pi-01a12625-8eb6-72a8-8fe0-31810bac58d6-reported-cost/run.json), [curation.json](https://capy.ai.id/evidence/curation/pi-01a12625-8eb6-72a8-8fe0-31810bac58d6-reported-cost.json), [transcript.json](https://capy.ai.id/evidence/runs/pi-01a12625-8eb6-72a8-8fe0-31810bac58d6-reported-cost/transcript.json), [response.txt](https://capy.ai.id/evidence/runs/pi-01a12625-8eb6-72a8-8fe0-31810bac58d6-reported-cost/response.txt), [HTML source](https://capy.ai.id/evidence/runs/pi-01a12625-8eb6-72a8-8fe0-31810bac58d6-reported-cost/output.html.txt)
- [gemini-3.8-flash](https://capy.ai.id/runs/pi-01a12606-d525-75a0-a529-3941c6b3563c-reported-cost/) (`pi-01a12606-d525-75a0-a529-3941c6b3563c-reported-cost`, google-vertex, tool-assisted, Pi (low effort); Model time: 317.707 s, Input tokens: 93,244, Cached tokens: 211,269, Output tokens: 39,515, Total tokens: 344,028, Cost (USD): $0.233959) — The preserved Pi session reports a total cost of US$0.233959425.
  Evidence: [run.json](https://capy.ai.id/evidence/runs/pi-01a12606-d525-75a0-a529-3941c6b3563c-reported-cost/run.json), [curation.json](https://capy.ai.id/evidence/curation/pi-01a12606-d525-75a0-a529-3941c6b3563c-reported-cost.json), [transcript.json](https://capy.ai.id/evidence/runs/pi-01a12606-d525-75a0-a529-3941c6b3563c-reported-cost/transcript.json), [response.txt](https://capy.ai.id/evidence/runs/pi-01a12606-d525-75a0-a529-3941c6b3563c-reported-cost/response.txt), [HTML source](https://capy.ai.id/evidence/runs/pi-01a12606-d525-75a0-a529-3941c6b3563c-reported-cost/output.html.txt)
- [gpt-6-luna](https://capy.ai.id/runs/pi-01a125fe-0142-763d-82c7-2f1f2e2650ee-reported-cost/) (`pi-01a125fe-0142-763d-82c7-2f1f2e2650ee-reported-cost`, openai-codex, one-shot, Pi (max effort); Model time: 81.426 s, Input tokens: 1,262, Cached tokens: 0, Output tokens: 11,195, Total tokens: 12,457, Cost (USD): $0.005724) — The preserved Pi session reports a total cost of US$0.005723700.
  Evidence: [run.json](https://capy.ai.id/evidence/runs/pi-01a125fe-0142-763d-82c7-2f1f2e2650ee-reported-cost/run.json), [curation.json](https://capy.ai.id/evidence/curation/pi-01a125fe-0142-763d-82c7-2f1f2e2650ee-reported-cost.json), [transcript.json](https://capy.ai.id/evidence/runs/pi-01a125fe-0142-763d-82c7-2f1f2e2650ee-reported-cost/transcript.json), [response.txt](https://capy.ai.id/evidence/runs/pi-01a125fe-0142-763d-82c7-2f1f2e2650ee-reported-cost/response.txt), [HTML source](https://capy.ai.id/evidence/runs/pi-01a125fe-0142-763d-82c7-2f1f2e2650ee-reported-cost/output.html.txt)
- [gpt-6-luna](https://capy.ai.id/runs/pi-01a125eb-9195-72d6-8736-a14e2a47f69a-reported-cost/) (`pi-01a125eb-9195-72d6-8736-a14e2a47f69a-reported-cost`, openai-codex, one-shot, Pi (xhigh effort); Model time: 52.654 s, Input tokens: 1,262, Cached tokens: 0, Output tokens: 6,733, Total tokens: 7,995, Cost (USD): $0.003493) — Self-contained HTML with inline CSS, SVG, and JavaScript animation.
  Evidence: [run.json](https://capy.ai.id/evidence/runs/pi-01a125eb-9195-72d6-8736-a14e2a47f69a-reported-cost/run.json), [curation.json](https://capy.ai.id/evidence/curation/pi-01a125eb-9195-72d6-8736-a14e2a47f69a-reported-cost.json), [transcript.json](https://capy.ai.id/evidence/runs/pi-01a125eb-9195-72d6-8736-a14e2a47f69a-reported-cost/transcript.json), [response.txt](https://capy.ai.id/evidence/runs/pi-01a125eb-9195-72d6-8736-a14e2a47f69a-reported-cost/response.txt), [HTML source](https://capy.ai.id/evidence/runs/pi-01a125eb-9195-72d6-8736-a14e2a47f69a-reported-cost/output.html.txt)
- [gpt-6-luna](https://capy.ai.id/runs/pi-01a125c6-e3d4-704e-977c-822bef6894cd-reported-cost/) (`pi-01a125c6-e3d4-704e-977c-822bef6894cd-reported-cost`, openai-codex, tool-assisted, Pi (high effort); Model time: 145.973 s, Input tokens: 8,137, Cached tokens: 0, Output tokens: 7,760, Total tokens: 15,897, Cost (USD): $0.004694) — Self-contained HTML with inline CSS and SVG animation.
  Evidence: [run.json](https://capy.ai.id/evidence/runs/pi-01a125c6-e3d4-704e-977c-822bef6894cd-reported-cost/run.json), [curation.json](https://capy.ai.id/evidence/curation/pi-01a125c6-e3d4-704e-977c-822bef6894cd-reported-cost.json), [transcript.json](https://capy.ai.id/evidence/runs/pi-01a125c6-e3d4-704e-977c-822bef6894cd-reported-cost/transcript.json), [response.txt](https://capy.ai.id/evidence/runs/pi-01a125c6-e3d4-704e-977c-822bef6894cd-reported-cost/response.txt), [HTML source](https://capy.ai.id/evidence/runs/pi-01a125c6-e3d4-704e-977c-822bef6894cd-reported-cost/output.html.txt)
- [gpt-6-luna](https://capy.ai.id/runs/pi-01a12554-7f7d-7778-8628-7f7f856c2e1a-reported-cost/) (`pi-01a12554-7f7d-7778-8628-7f7f856c2e1a-reported-cost`, openai-codex, tool-assisted, Pi (medium effort); Model time: 99.841 s, Input tokens: 5,236, Cached tokens: 0, Output tokens: 5,361, Total tokens: 10,597, Cost (USD): $0.003204) — Self-contained HTML with inline CSS and CSS keyframe animation.
  Evidence: [run.json](https://capy.ai.id/evidence/runs/pi-01a12554-7f7d-7778-8628-7f7f856c2e1a-reported-cost/run.json), [curation.json](https://capy.ai.id/evidence/curation/pi-01a12554-7f7d-7778-8628-7f7f856c2e1a-reported-cost.json), [transcript.json](https://capy.ai.id/evidence/runs/pi-01a12554-7f7d-7778-8628-7f7f856c2e1a-reported-cost/transcript.json), [response.txt](https://capy.ai.id/evidence/runs/pi-01a12554-7f7d-7778-8628-7f7f856c2e1a-reported-cost/response.txt), [HTML source](https://capy.ai.id/evidence/runs/pi-01a12554-7f7d-7778-8628-7f7f856c2e1a-reported-cost/output.html.txt)
- [gpt-6-luna](https://capy.ai.id/runs/pi-01a123fb-b53f-727e-b6e5-c487a4e04628-metrics-reported-cost/) (`pi-01a123fb-b53f-727e-b6e5-c487a4e04628-metrics-reported-cost`, openai-codex, one-shot, Pi (low effort); Model time: 17.222 s, Input tokens: 1,262, Cached tokens: 0, Output tokens: 2,464, Total tokens: 3,726, Cost (USD): $0.001358) — Collected with Pi 1.1.0, openai-codex/gpt-6-luna, thinking level low.
  Evidence: [run.json](https://capy.ai.id/evidence/runs/pi-01a123fb-b53f-727e-b6e5-c487a4e04628-metrics-reported-cost/run.json), [curation.json](https://capy.ai.id/evidence/curation/pi-01a123fb-b53f-727e-b6e5-c487a4e04628-metrics-reported-cost.json), [transcript.json](https://capy.ai.id/evidence/runs/pi-01a123fb-b53f-727e-b6e5-c487a4e04628-metrics-reported-cost/transcript.json), [response.txt](https://capy.ai.id/evidence/runs/pi-01a123fb-b53f-727e-b6e5-c487a4e04628-metrics-reported-cost/response.txt), [HTML source](https://capy.ai.id/evidence/runs/pi-01a123fb-b53f-727e-b6e5-c487a4e04628-metrics-reported-cost/output.html.txt)