[object Object]

← back to Model Arena

auto-data-snapshot: 2026-08-23T17:36:09 (8 data files) — data/artifacts/34f7125fa421/claude-code.png data/artifacts/34f7125fa421/qwen25-7b.png data/artifacts/34f7125fa421/qwen3-14b.png data/artifacts/8149a54eaade/qwen25-7b.png data/artifacts/8149a54eaade/qwen3-14b.png

b07fd902543a910a0b7e57d1d446cb3c49503841 · 2026-08-23 17:36:23 -0700 · auto-commit-fleet

Files touched

Diff

commit b07fd902543a910a0b7e57d1d446cb3c49503841
Author: auto-commit-fleet <steve@designerwallcoverings.com>
Date:   Sun Aug 23 17:36:23 2026 -0700

    auto-data-snapshot: 2026-08-23T17:36:09 (8 data files) — data/artifacts/34f7125fa421/claude-code.png data/artifacts/34f7125fa421/qwen25-7b.png data/artifacts/34f7125fa421/qwen3-14b.png data/artifacts/8149a54eaade/qwen25-7b.png data/artifacts/8149a54eaade/qwen3-14b.png
---
 data/artifacts/34f7125fa421/claude-code.png | Bin 0 -> 247331 bytes
 data/artifacts/34f7125fa421/qwen25-7b.png   | Bin 0 -> 68951 bytes
 data/artifacts/34f7125fa421/qwen3-14b.png   | Bin 0 -> 2724 bytes
 data/artifacts/8149a54eaade/qwen25-7b.png   | Bin 0 -> 33354 bytes
 data/artifacts/8149a54eaade/qwen3-14b.png   | Bin 0 -> 95701 bytes
 data/artifacts/9000631deff3/qwen3-14b.png   | Bin 3310 -> 61503 bytes
 data/challenges.json                        | 158 +++++++++++++++++++++++++---
 data/votes.jsonl                            |   1 +
 8 files changed, 147 insertions(+), 12 deletions(-)

diff --git a/data/artifacts/34f7125fa421/claude-code.png b/data/artifacts/34f7125fa421/claude-code.png
new file mode 100644
index 0000000..4a0ec57
Binary files /dev/null and b/data/artifacts/34f7125fa421/claude-code.png differ
diff --git a/data/artifacts/34f7125fa421/qwen25-7b.png b/data/artifacts/34f7125fa421/qwen25-7b.png
new file mode 100644
index 0000000..580c202
Binary files /dev/null and b/data/artifacts/34f7125fa421/qwen25-7b.png differ
diff --git a/data/artifacts/34f7125fa421/qwen3-14b.png b/data/artifacts/34f7125fa421/qwen3-14b.png
new file mode 100644
index 0000000..ad57eb4
Binary files /dev/null and b/data/artifacts/34f7125fa421/qwen3-14b.png differ
diff --git a/data/artifacts/8149a54eaade/qwen25-7b.png b/data/artifacts/8149a54eaade/qwen25-7b.png
new file mode 100644
index 0000000..73da931
Binary files /dev/null and b/data/artifacts/8149a54eaade/qwen25-7b.png differ
diff --git a/data/artifacts/8149a54eaade/qwen3-14b.png b/data/artifacts/8149a54eaade/qwen3-14b.png
new file mode 100644
index 0000000..6659e57
Binary files /dev/null and b/data/artifacts/8149a54eaade/qwen3-14b.png differ
diff --git a/data/artifacts/9000631deff3/qwen3-14b.png b/data/artifacts/9000631deff3/qwen3-14b.png
index bd3cc02..2bd30f9 100644
Binary files a/data/artifacts/9000631deff3/qwen3-14b.png and b/data/artifacts/9000631deff3/qwen3-14b.png differ
diff --git a/data/challenges.json b/data/challenges.json
index 26f9996..aff2a47 100644
--- a/data/challenges.json
+++ b/data/challenges.json
@@ -32528,12 +32528,12 @@
         "model": "qwen3-14b",
         "status": "done",
         "error": null,
-        "seconds": 21,
+        "seconds": 26,
         "cost": 0,
-        "started_at": "2026-08-20T14:15:05.454Z",
-        "finished_at": "2026-08-20T14:15:26.839Z",
-        "queued_at": "2026-08-20T14:15:05.435Z",
-        "bytes": 4098,
+        "started_at": "2026-08-24T00:22:18.933Z",
+        "finished_at": "2026-08-24T00:22:45.328Z",
+        "queued_at": "2026-08-24T00:22:18.927Z",
+        "bytes": 3950,
         "thumb": true,
         "aiScore": 5.8,
         "aiReason": "The model successfully implements interactive fireworks and basic physics effects but lacks the auto-show mode.",
@@ -32659,18 +32659,18 @@
     "category": "Custom",
     "designTools": false,
     "created_at": "2026-08-22T14:15:01.807Z",
-    "winner": null,
+    "winner": "qwen3-14b",
     "runs": [
       {
         "model": "qwen3-14b",
         "status": "done",
         "error": null,
-        "seconds": 23,
+        "seconds": 14,
         "cost": 0,
-        "started_at": "2026-08-22T14:15:01.833Z",
-        "finished_at": "2026-08-22T14:15:24.650Z",
-        "queued_at": "2026-08-22T14:15:01.814Z",
-        "bytes": 3920,
+        "started_at": "2026-08-24T00:20:17.102Z",
+        "finished_at": "2026-08-24T00:20:31.258Z",
+        "queued_at": "2026-08-24T00:20:17.097Z",
+        "bytes": 2703,
         "thumb": true
       },
       {
@@ -32715,7 +32715,8 @@
     ],
     "judging": false,
     "aiPick": "qwen25-7b",
-    "judged_at": "2026-08-22T14:15:40.056Z"
+    "judged_at": "2026-08-22T14:15:40.056Z",
+    "voted_at": "2026-08-24T00:20:59.395Z"
   },
   {
     "id": "4a268e78cf14",
@@ -32781,5 +32782,138 @@
     "judging": false,
     "aiPick": "qwen25-7b",
     "judged_at": "2026-08-23T14:15:43.183Z"
+  },
+  {
+    "id": "34f7125fa421",
+    "title": "Dragon Realm",
+    "prompt": "Build an animated dragon realm scene: a dragon flying over mountains with snowfall, dynamic lighting, and at least one interactive element (click to make the dragon breathe fire). Visual quality matters most.",
+    "category": "Games",
+    "designTools": false,
+    "created_at": "2026-08-24T00:24:06.644Z",
+    "winner": null,
+    "runs": [
+      {
+        "model": "qwen3-14b",
+        "status": "done",
+        "error": null,
+        "seconds": 38,
+        "cost": 0,
+        "started_at": "2026-08-24T00:24:06.664Z",
+        "finished_at": "2026-08-24T00:24:44.769Z",
+        "queued_at": "2026-08-24T00:24:06.650Z",
+        "bytes": 5452,
+        "thumb": true,
+        "aiScore": 4.8,
+        "aiReason": "The image is mostly black, indicating it's either not rendered or broken.",
+        "aiScores": {
+          "qwen2.5vl:7b": 5,
+          "minicpm-v:latest": 4.5
+        },
+        "aiSpread": 0.5
+      },
+      {
+        "model": "qwen25-7b",
+        "status": "done",
+        "error": null,
+        "seconds": 26,
+        "cost": 0,
+        "started_at": "2026-08-24T00:24:06.669Z",
+        "finished_at": "2026-08-24T00:24:32.492Z",
+        "queued_at": "2026-08-24T00:24:06.655Z",
+        "bytes": 2898,
+        "thumb": true,
+        "aiScore": 5,
+        "aiReason": "The image lacks dynamic lighting and the dragon is not flying over mountains.",
+        "aiScores": {
+          "qwen2.5vl:7b": 6,
+          "minicpm-v:latest": 4
+        },
+        "aiSpread": 2
+      },
+      {
+        "model": "claude-code",
+        "status": "done",
+        "error": null,
+        "seconds": 101,
+        "cost": 0,
+        "started_at": "2026-08-24T00:24:06.674Z",
+        "finished_at": "2026-08-24T00:25:47.439Z",
+        "queued_at": "2026-08-24T00:24:06.659Z",
+        "bytes": 13744,
+        "thumb": true,
+        "aiScore": 8.3,
+        "aiReason": "The scene is visually engaging and fulfills all the challenge requirements effectively.",
+        "aiScores": {
+          "qwen2.5vl:7b": 9,
+          "minicpm-v:latest": 7.5
+        },
+        "aiSpread": 1.5
+      }
+    ],
+    "judging": false,
+    "aiPick": "claude-code",
+    "judged_at": "2026-08-24T00:26:10.253Z"
+  },
+  {
+    "id": "8149a54eaade",
+    "title": "Ultimate battle",
+    "prompt": "Create a visually spectacular 2026 interactive website called “MODEL WARS” where leading AI models battle head-to-head in a medieval tournament.\n\nCore Concept\n\nTurn AI model comparisons into an elaborate medieval tournament, combining jousting, castles, knights, banners, tournaments, and arena combat with sophisticated 2026 liquid-glass graphics, fluid animation, particles, reflections, and cinematic lighting.\n\nEach AI model becomes a medieval champion:\n\n* OpenAI / GPT — The Emerald Knight\n* Claude — The Golden Scholar\n* Gemini — The Celestial Mage\n* Grok — The Black Knight\n* DeepSeek — The Eastern Strategist\n* Llama — The Crimson Ranger\n* Allow new models to be added dynamically.\n\nDo not simply create a benchmark dashboard. Make the benchmarks feel like competitive games people want to watch.\n\nTHE TOURNAMENT\n\nUsers select two models and choose a battle.\n\n⚔️ JOUST\n\nTwo armored knights charge toward one another on horseback.\n\nBehind the animation, both models receive the exact same prompt.\n\nTheir responses are evaluated for:\n\n* Accuracy\n* Reasoning\n* Speed\n* Instruction following\n* Hallucinations\n* Completeness\n\nEach category contributes to the strength of the jousting strike.\n\nThe winning knight knocks the opponent from the horse.\n\nDisplay the actual scoring immediately afterward.\n\n🏹 ARCHERY\n\nModels receive precision questions.\n\nEach correct answer moves an arrow closer to the bullseye.\n\nAccuracy determines placement.\n\nAnimate arrows flying toward large medieval targets.\n\n🏰 CASTLE SIEGE\n\nUse difficult multi-step reasoning problems.\n\nEach successful reasoning stage destroys part of the opponent’s castle:\n\n* Gate\n* Walls\n* Towers\n* Keep\n\nShow trebuchets, flaming projectiles, smoke, debris, banners and collapsing masonry.\n\n⚔️ DUEL\n\nModels solve coding, mathematics or logic challenges.\n\nRepresent progress through sword combat.\n\nCorrect reasoning = successful attack.\n\nErrors = blocked attacks or lost health.\n\n🐉 DRAGON CHALLENGE\n\nGive both models an extremely difficult problem.\n\nThe dragon represents the problem itself.\n\nModels aren’t attacking each other—they are competing to defeat the same dragon first.\n\nShow the dragon’s health decreasing as each model completes valid reasoning stages.\n\nLIVE MODEL ARENA\n\nThe user can enter ANY prompt.\n\nChoose:\n\nModel A vs Model B\n\nThen click:\n\nBEGIN BATTLE\n\nSend the identical prompt to both model APIs simultaneously.\n\nNever fake model responses or benchmark results.\n\nShow:\n\n* Response\n* Time to first token\n* Total generation time\n* Token usage\n* Estimated API cost\n* Judge score\n* Winner\n\nProvide a VIEW THE EVIDENCE button exposing the original outputs and complete scoring methodology.\n\nAI JUDGES\n\nUse multiple independent judges where possible.\n\nScore each answer from 0–100 for:\n\n* Correctness\n* Reasoning\n* Relevance\n* Completeness\n* Instruction adherence\n* Hallucination risk\n\nClearly distinguish objective measurements from AI-judge opinions.\n\nAllow HUMAN VOTE.\n\nMaintain separate:\n\nAI SCORE\n\nand\n\nPEOPLE’S CHOICE\n\nMEDIEVAL ELO SYSTEM\n\nEvery model has a persistent tournament rating.\n\nExample:\n\nGPT — 1847\nClaude — 1821\nGemini — 1789\nGrok — 1744\n\nRatings rise and fall after battles.\n\nCreate:\n\nTHE KING’S TABLE\n\nas the global leaderboard.\n\nInclude:\n\n* Rank\n* Model\n* Provider\n* ELO\n* Wins\n* Losses\n* Draws\n* Win %\n* Current streak\n* Strongest category\n* Average response time\n* Cost efficiency\n\nGive the #1 model a subtly animated crown.\n\n2026 VISUAL LANGUAGE\n\nThe interface should look like:\n\nMedieval tournament × AAA fantasy game × futuristic liquid-glass operating system.\n\nUse:\n\n* Liquid glass panels\n* Transparent refractive surfaces\n* Fluid distortion\n* Metallic armor\n* Gold and silver trim\n* Deep blacks\n* Volumetric lighting\n* Fire particles\n* Embers\n* Smoke\n* Reflective water\n* Animated banners\n* Medieval typography for major headings\n* Modern typography for data\n* Cinematic depth of field\n* Parallax castles\n* Interactive torchlight\n* Dynamic shadows\n\nAvoid a cheap cartoon aesthetic.\n\nThe medieval ",
+    "category": "Games",
+    "designTools": true,
+    "created_at": "2026-08-24T00:28:44.853Z",
+    "winner": null,
+    "runs": [
+      {
+        "model": "qwen3-14b",
+        "status": "done",
+        "error": null,
+        "seconds": 51,
+        "cost": 0,
+        "started_at": "2026-08-24T00:28:44.876Z",
+        "finished_at": "2026-08-24T00:29:35.967Z",
+        "queued_at": "2026-08-24T00:28:44.859Z",
+        "bytes": 6288,
+        "thumb": true,
+        "aiScore": 1.8,
+        "aiReason": "The image is completely black and does not provide any visual content or information related to the challenge.",
+        "aiScores": {
+          "qwen2.5vl:7b": 0,
+          "minicpm-v:latest": 3.5
+        },
+        "aiSpread": 3.5
+      },
+      {
+        "model": "qwen25-7b",
+        "status": "done",
+        "error": null,
+        "seconds": 35,
+        "cost": 0,
+        "started_at": "2026-08-24T00:28:44.881Z",
+        "finished_at": "2026-08-24T00:29:20.288Z",
+        "queued_at": "2026-08-24T00:28:44.864Z",
+        "bytes": 4541,
+        "thumb": true,
+        "aiScore": 7.3,
+        "aiReason": "The image shows a basic selection interface for AI models in a medieval-themed battle game. While it captures the essence of the challenge by presenting models as characters and providing options to s",
+        "aiScores": {
+          "qwen2.5vl:7b": 7,
+          "minicpm-v:latest": 7.5
+        },
+        "aiSpread": 0.5
+      },
+      {
+        "model": "claude-code",
+        "status": "error",
+        "error": "claude CLI: Command failed: /Users/macstudio3/.npm-global/bin/claude --model opus --append-system-prompt You are a headless code generator in a build arena. Respond with ON",
+        "seconds": 301,
+        "cost": null,
+        "started_at": "2026-08-24T00:28:44.888Z",
+        "finished_at": "2026-08-24T00:33:45.503Z",
+        "queued_at": "2026-08-24T00:28:44.870Z"
+      }
+    ],
+    "judging": false,
+    "aiPick": "qwen25-7b",
+    "judged_at": "2026-08-24T00:34:04.020Z"
   }
 ]
\ No newline at end of file
diff --git a/data/votes.jsonl b/data/votes.jsonl
index c1d2baf..12157b2 100644
--- a/data/votes.jsonl
+++ b/data/votes.jsonl
@@ -54,3 +54,4 @@
 {"ts":"2026-07-25T06:34:44.276Z","challenge":"0cb3fc06b069","title":"Free-Roster — Particle Text","winner":"claude-code","field":["qwen3-14b","gemma3-12b","hermes3-8b","qwen25-7b","hf-qwen-coder-32b","claude-code"]}
 {"ts":"2026-07-25T15:29:40.133Z","challenge":"7c8685c3058e","title":"Designer Wallcoverings — Colorway Swatch Match Game","winner":"grok","field":["qwen3-14b","gemma3-12b","hermes3-8b","qwen25-7b","hf-qwen-coder-32b","claude-code","kimi","gpt","grok"]}
 {"ts":"2026-07-26T00:19:08.660Z","challenge":"258516d076e6","title":"Designer Wallcoverings — Shop by Color Wheel","winner":"claude-code","field":["qwen3-14b","gemma3-12b","hermes3-8b","qwen25-7b","hf-qwen-coder-32b","claude-code","kimi","gpt","grok"]}
+{"ts":"2026-08-24T00:20:59.395Z","challenge":"9000631deff3","title":"Daily: Bouncing Balls","winner":"qwen3-14b","field":["qwen3-14b","qwen25-7b"]}

← e2ec462 auto-data-snapshot: 2026-08-23T07:31:56 (4 data files) — dat  ·  back to Model Arena  ·  auto-data-snapshot: 2026-08-23T18:39:41 (4 data files) — dat 5ddfae1 →