BenchLeader

Design Arena Code leaderboard

As of 21 Nov 2026, the top score on Design Arena Code is 1406 by GPT-6.2 Sol (OpenAI), ahead of Claude Fable 5.2 at 1392, across 188 model configurations with a published result (Design Arena, result dated 21 Nov 2026). Two models write front-end code (HTML) for the same text prompt in any of Design Arena's code categories, such as websites, UI components, games, data visualisations and 3D scenes; visitors vote for the better result and the votes become one Bradley-Terry rating across those categories. Models with fewer than 200 battles are left out. Design Arena Code is shown for reference only and does not count toward the BenchLeader Index.

Key facts (as of 21 Nov 2026)

  • GPT-6.2 Sol holds the top Design Arena Code score at 1406 (Design Arena, 21 Nov 2026).
  • Claude Fable 5.2 is second on Design Arena Code at 1392 (Design Arena, 21 Nov 2026).
  • Design Arena Code: Two models write front-end code (HTML) for the same text prompt in any of Design Arena's code categories, such as websites, UI components, games, data visualisations and 3D scenes; visitors vote for the better result and the votes become one Bradley-Terry rating across those categories. Models with fewer than 200 battles are left out (Design Arena).
  • 188 model configurations have a published Design Arena Code result on BenchLeader (BenchLeader, 21 Nov 2026).
  • Design Arena Code does not feed the BenchLeader Index; it is shown for reference (BenchLeader methodology).

Two models write front-end code (HTML) for the same text prompt in any of Design Arena's code categories, such as websites, UI components, games, data visualisations and 3D scenes; visitors vote for the better result and the votes become one Bradley-Terry rating across those categories. Models with fewer than 200 battles are left out.

Published by
Design Arena
Category
Coding
Index weight
Reference only
Models
188
Data as of
21 Nov 2026

© Arcada Labs. Design Arena (https://www.designarena.ai/leaderboard). Ratings read daily from the public leaderboard and shown with credit and a link; not endorsed by Arcada Labs.

188 of 188
#
1NewGPT-6.2 Sol1406–2026-11-21
2Claude Fable 5.21392–2026-11-21
3NewGemini 4 Neon1390–2026-11-21
4NewGPT-6.2 Astra1390–2026-11-21
5GPT-6 Astraxhigh138964.92026-11-21
6Muse Wondermax138460.12026-11-21
7Claude Opus 5.5138162.62026-11-21
8Inkling 2 Small137757.22026-11-21
9GPT-6 Astramax137466.82026-11-21
10GPT-6.1 Sol1352–2026-11-21
11Claude Sonnet 5.51352–2026-11-21
12Gemini 4 Carbon1350–2026-11-21
13Kimi K3134560.62026-11-21
14Muse Spark 1.3max134560.12026-11-21
15Muse Spark 1.3xhigh134359.02026-11-21
16Kimi K3.1 Pro134160.42026-11-21
17Muse Ember1340–2026-11-21
18GPT-5.6 Solxhigh133163.72026-11-21
19Gemini 4 Hydrogen1329–2026-11-21
20Claude Fable 5.1132266.92026-11-21
21Gemini 4 Argon131963.32026-11-21
22Claude Opus 5130963.22026-11-21
23MiMo-V2.6-Pro130857.02026-11-21
24DeepSeek V4.1 Flash1301–2026-11-21
25GPT-5.6 Solmedium130061.12026-11-21
26Muse Spark 1.21298–2026-11-21
27Ling 3.1 Flash129550.72026-11-21
28Claude Fable 5129365.12026-11-21
29GPT-6.1 Terra1293–2026-11-21
30Gemini 3.7 Flash1290–2026-11-21
31Muse Spark Openxhigh128955.32026-11-21
32MiniMax M3.1 Flash128953.12026-11-21
33GPT-6 Sol1287–2026-11-21
34Gemini 3.8 Flash128555.72026-11-21
35DeepSeek V4.1 Flash 10171281–2026-11-21
36Qwen3.8 Max127960.82026-11-21
37GLM 5.31275–2026-11-21
38Claude Haiku 5.51273–2026-11-21
39Gemini 3.6 Flash127248.92026-11-21
40Grok 4.61272–2026-11-21
41Claude Opus 4.6127060.72026-11-21
42GLM 5.2126851.82026-11-21
43Claude Opus 4.6thinking1267–2026-11-21
44Claude Sonnet 5126052.52026-11-21
45GLM 5.3 Flash125957.52026-11-21
46Grok 4.71259–2026-11-21
47Muse Spark 1.1125861.62026-11-21
48Claude Sonnet 4.6125858.32026-11-21
49Qwen3.7 Max125759.82026-11-21
50Grok 4.5125658.92026-11-21
51MiMo-V2.5-Pro125555.52026-11-21
52Kimi K2.6125455.02026-11-21
53GPT-6 Luna1253–2026-11-21
54GLM 5.1125255.22026-11-21
55Qwen3.7 Plus125157.52026-11-21
56Gemini 3.5 Flash1249–2026-11-21
57Hy4124756.02026-11-21
58Mistral Large 4124751.82026-11-21
59MiMo-V2.5124456.32026-11-21
60GLM 5 Turbo1244–2026-11-21

AI coding benchmark leaderboard · All LLM benchmarks

Cite as: BenchLeader, “Design Arena Code leaderboard”, data as of 21 Nov 2026, https://www.benchleader.com/benchmarks/designarena_code.

Design Arena Code: questions

What does Design Arena Code measure?
Two models write front-end code (HTML) for the same text prompt in any of Design Arena's code categories, such as websites, UI components, games, data visualisations and 3D scenes; visitors vote for the better result and the votes become one Bradley-Terry rating across those categories. Models with fewer than 200 battles are left out. Scores are reported in Elo-style ratings from pairwise comparisons; higher is better.
What is the top score on Design Arena Code?
As of 21 Nov 2026, the highest score on Design Arena Code is 1406 by GPT-6.2 Sol (result dated 21 Nov 2026), ahead of Claude Fable 5.2 at 1392. Scores come from Design Arena and are re-read daily.
Which AI model leads Design Arena Code?
GPT-6.2 Sol leads Design Arena Code with 1406 as of 21 Nov 2026 (result dated 21 Nov 2026), ahead of Claude Fable 5.2 at 1392.
How many models have Design Arena Code results?
188 model configurations have a Design Arena Code result on BenchLeader, all taken from Design Arena.
Who runs Design Arena Code and how often is it updated?
Design Arena Code is published by Design Arena. BenchLeader re-reads the published results once a day and records the date each result was published.
Does Design Arena Code count toward the BenchLeader Index?
No. Design Arena Code is shown for reference but left out of the composite index.