BenchLeader
Menu
Best LLMsChoose a modelCoding plansLeaderboardsModelsBenchmarksProvidersReleases12

Design Arena Agentic leaderboard

As of 9 Nov 2026, the top score on Design Arena Agentic is 1461 by Claude Opus 5.5 (Anthropic), ahead of GPT-6.1 Sol at 1425, across 77 model configurations with a published result (Design Arena, result dated 9 Nov 2026). Models run as agents with tools to build web apps, slide decks, games and 3D scenes; visitors compare two finished results and vote. Design Arena Agentic is shown for reference only and does not count toward the BenchLeader Index.

Key facts (as of 9 Nov 2026)

  • Claude Opus 5.5 holds the top Design Arena Agentic score at 1461 (Design Arena, 9 Nov 2026).
  • GPT-6.1 Sol is second on Design Arena Agentic at 1425 (Design Arena, 9 Nov 2026).
  • Design Arena Agentic: Models working as agents build web apps, slide decks, games and 3D scenes; visitors vote between two results and the votes become a Bradley-Terry rating, overall across those builds. Models with fewer than 200 battles are left out (Design Arena).
  • 77 model configurations have a published Design Arena Agentic result on BenchLeader (BenchLeader, 9 Nov 2026).
  • Design Arena Agentic does not feed the BenchLeader Index; it is shown for reference (BenchLeader methodology).

Models working as agents build web apps, slide decks, games and 3D scenes; visitors vote between two results and the votes become a Bradley-Terry rating, overall across those builds. Models with fewer than 200 battles are left out.

Published by
Design Arena
Category
Agents & tools
Index weight
Reference only
Models
77
Data as of
9 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.

What the test looks like

Models run as agents with tools to build web apps, slide decks, games and 3D scenes; visitors compare two finished results and vote.

How it is scored

Bradley-Terry rating from the pairwise votes, overall across the agent arenas. Effort settings are listed separately where Design Arena runs them.

What to keep in mind

Each model runs in Design Arena's harness, so a model's own coding agent may do better or worse. Fewer battles per model than the website board. Kept out of the index.

Top 15
  1. 1Claude Opus 5.5 (high)1461
  2. 2GPT-6.1 Sol (max)1425
  3. 3GPT-6 Astra (max)1393
  4. 4Claude Opus 5.51334
  5. 5Claude Fable 5.11332
  6. 6DeepSeek V4.1 Flash 10171304
  7. 7GPT-6 Astra (xhigh)1301
  8. 8Kimi K3.1 Pro1284
  9. 9Claude Sonnet 5.51283
  10. 10Gemini 4 Argon1280
  11. 11Muse Spark 1.3 (max)1278
  12. 12Inkling 2 Small1272
  13. 13Claude Fable 51268
  14. 14Claude Opus 51266
  15. 15MiMo-V2.6-Pro1266
77 of 77
#
1Claude Opus 5.5high146165.42026-11-09
2GPT-6.1 Solmax142567.02026-11-09
3GPT-6 Astramax139367.42026-11-09
4Claude Opus 5.5133463.22026-11-09
5Claude Fable 5.1133266.02026-11-09
6DeepSeek V4.1 Flash 10171304–2026-11-09
7GPT-6 Astraxhigh130166.22026-11-09
8Kimi K3.1 Pro128461.22026-11-09
9Claude Sonnet 5.51283–2026-11-09
10Gemini 4 Argon128063.32026-11-09
11Muse Spark 1.3max127861.72026-11-09
12NewInkling 2 Small127258.32026-11-09
13Claude Fable 5126866.62026-11-09
14Claude Opus 5126665.22026-11-09
15MiMo-V2.6-Pro126657.12026-11-09
16Muse Spark 1.3xhigh125958.12026-11-09
17Kimi K3125759.52026-11-09
18Qwen3.8 Maxmax125558.92026-11-09
19GPT-6.1 Sol125062.22026-11-09
20Grok 4.71239–2026-11-09
21MiniMax M3.1 Flash123853.92026-11-09
22GPT-5.6 Solxhigh123664.22026-11-09
23Grok 4.6123356.82026-11-09
24GLM 5.31230–2026-11-09
25Claude Opus 4.8122160.12026-11-09
26Gemini 3.8 Flash122054.32026-11-09
27Gemini 3.7 Flash1218–2026-11-09
28Claude Sonnet 5121756.02026-11-09
29Grok 4.5121559.92026-11-09
30GPT-6 Sol1212–2026-11-09
31Ling 3.1 Flash120050.82026-11-09
32Claude Opus 4.6119262.32026-11-09
33GLM 5.2119051.52026-11-09
34GPT-6.1 Terra1190–2026-11-09
35GPT-5.6 Solmedium118761.42026-11-09
36MiniMax M3118256.32026-11-09
37Claude Sonnet 4.6118158.22026-11-09
38Muse Spark 1.21181–2026-11-09
39Qwen3.7max117460.32026-11-09
40Gemini 3.6 Flash117149.62026-11-09
41Gemini 3.5 Flash116848.82026-11-09
42Inkling116847.72026-11-09
43Muse Spark 1.1116762.32026-11-09
44Muse Wondermax116661.92026-11-09
45GLM 5.1115755.82026-11-09
46Kimi K2.6115355.22026-11-09
47Kimi K2.7 Code115354.22026-11-09
48Claude Opus 4.5115157.92026-11-09
49GPT-6 Luna1149–2026-11-09
50GPT-5.6 Luna114247.72026-11-09
51Kimi K2.5thinking111158.82026-11-09
52GLM-5111154.92026-11-09
53GPT-5.5110861.72026-11-09
54GLM 5V Turbo110555.82026-11-09
55Grok 4.20thinking108958.42026-11-09
56GPT-5.6 Terra108651.72026-11-09
57Gemini 3.1 Pro108360.62026-11-09
58Solar Pro 4107949.52026-11-09
59Gemini 3 Flash107656.92026-11-09
60Gemini 3 Pro107560.32026-11-09

More agents & tools benchmarks · All LLM benchmarks

Related benchmarks

Top head-to-heads on Design Arena Agentic

Cite as: BenchLeader, “Design Arena Agentic leaderboard”, data as of 9 Nov 2026, https://www.benchleader.com/benchmarks/designarena_agentic.

Design Arena Agentic: questions

What does Design Arena Agentic measure?
Models run as agents with tools to build web apps, slide decks, games and 3D scenes; visitors compare two finished results and vote. Scores are reported in Elo-style ratings from pairwise comparisons; higher is better.
What is the top score on Design Arena Agentic?
As of 9 Nov 2026, the highest score on Design Arena Agentic is 1461 by Claude Opus 5.5 (result dated 9 Nov 2026), ahead of GPT-6.1 Sol at 1425. Scores come from Design Arena and are re-read daily.
Which AI model leads Design Arena Agentic?
Claude Opus 5.5 leads Design Arena Agentic with 1461 as of 9 Nov 2026 (result dated 9 Nov 2026), ahead of GPT-6.1 Sol at 1425.
How many models have Design Arena Agentic results?
77 model configurations have a Design Arena Agentic result on BenchLeader, all taken from Design Arena.
Who runs Design Arena Agentic and how often is it updated?
Design Arena Agentic 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 Agentic count toward the BenchLeader Index?
No. Design Arena Agentic is shown for reference but left out of the composite index.