BenchLeader
Menu
Best LLMsChoose a modelLeaderboardsModelsBenchmarksProvidersReleases15

LiveBench Design

LiveBench 2.0 design suite.

As of 21 Oct 2026, Claude Opus 5.5 leads LiveBench Design on BenchLeader with 86.9%, ahead of Claude Sonnet 5.5 at 85.0%, across 48 model configurations with a published result.

Published by
LiveBench
Category
Multimodal
Index weight
Reference only
Models
48
Data as of
21 Oct 2026

LiveBench (livebench.ai), White et al. 2024.

What the test looks like

LiveBench's design suite: layout and visual design tasks graded against a fixed rubric.

How it is scored

Mean task score for the suite, 0–100, from a single run.

What to keep in mind

A design rubric is partly subjective even when automated; treat it as one suite among many.

Top 15
  1. 1Claude Opus 5.586.9%
  2. 2Claude Sonnet 5.585.0%
  3. 3Claude Fable 584.7%
  4. 4Claude Fable 5.183.4%
  5. 5DeepSeek V4.1 Flash 101783.4%
  6. 6Claude Haiku 5.583.1%
  7. 7Claude Opus 583.0%
  8. 8GPT-6 Luna82.5%
  9. 9GPT-6 Sol82.2%
  10. 10GPT-6.1 Sol82.1%
  11. 11GPT-6 Astra82.0%
  12. 12Claude Sonnet 582.0%
  13. 13GPT-5.6 Sol81.6%
  14. 14Grok 4.679.8%
  15. 15DeepSeek V4.1 Flash79.2%
48 of 48
#
1Claude Opus 5.586.9%61.92026-10-14
2Claude Sonnet 5.585.0%–2026-10-14
3Claude Fable 584.7%67.12026-10-14
4Claude Fable 5.183.4%66.22026-10-14
5NewDeepSeek V4.1 Flash 101783.4%–2026-10-14
6Claude Haiku 5.583.1%–2026-10-14
7Claude Opus 583.0%65.42026-10-14
8GPT-6 Luna82.5%–2026-10-14
9GPT-6 Sol82.2%–2026-10-14
10GPT-6.1 Sol82.1%63.32026-10-14
11GPT-6 Astra82.0%67.92026-10-14
12Claude Sonnet 582.0%56.02026-10-14
13GPT-5.6 Sol81.6%55.92026-10-14
14Grok 4.679.8%57.82026-10-14
15DeepSeek V4.1 Flash79.2%–2026-10-14
16Claude Opus 4.878.8%60.12026-10-14
17Qwen3.8 Maxmax78.6%63.32026-10-14
18MiMo-V2.6-Pro78.5%64.12026-10-14
19Grok 4.578.1%58.92026-10-14
20Grok 4.778.0%–2026-10-14
21GLM 5.377.3%–2026-10-14
22GLM 5.3 Flash77.2%62.72026-10-14
23Kimi K377.2%59.12026-10-14
24Step 5 Preview76.9%60.62026-10-14
25GPT-5.6 Terra76.0%52.62026-10-14
26GPT-5.6 Luna76.0%47.92026-10-14
27Qwen3.8 27B75.9%54.52026-10-14
28Muse Wonder75.1%63.62026-10-14
29Muse Spark 1.374.8%58.32026-10-14
30NewMiniMax-M3.1-Flash74.0%61.92026-10-14
31Gemini 4 Argon73.7%63.32026-10-14
32Gemini 3.7 Flash73.2%53.22026-10-14
33Qwen3.8 Flash72.8%–2026-10-14
34Gemini 3.8 Flash72.7%55.62026-10-14
35Claude Sonnet 4.671.8%57.12026-10-14
36Deepseek v4 Flash Vision70.7%–2026-10-14
37GPT-5.570.3%60.22026-10-14
38NewBeam 1.067.9%54.62026-10-14
39Muse Spark 1.267.1%–2026-10-14
40MiniMax M366.2%56.52026-10-14
41Mistral Large 465.6%56.72026-10-14
42Gemini 3.1 Pro61.6%60.72026-10-14
43Nemotron Super VL 3.560.7%–2026-10-14
44NewMuse Spark Open60.3%–2026-10-14
45Claude Haiku 4.559.9%45.32026-10-14
46Nemotron Ultra 358.6%–2026-10-14
47Gemma 4 31B58.4%48.32026-10-14
48Nemotron Lightning 3.543.9%–2026-10-14

More multimodal benchmarks · All LLM benchmarks

Cite as: BenchLeader, “LiveBench Design leaderboard”, https://www.benchleader.com/benchmarks/livebench_design, data as of 21 Oct 2026.

LiveBench Design: questions

What does LiveBench Design measure?
LiveBench's design suite: layout and visual design tasks graded against a fixed rubric. Scores are reported in percent of tasks solved; higher is better.
Which AI model leads LiveBench Design?
Claude Opus 5.5 leads LiveBench Design with 86.9% as of 21 Oct 2026, ahead of Claude Sonnet 5.5 at 85.0%.
How many models have LiveBench Design results?
48 model configurations have a LiveBench Design result on BenchLeader, all taken from LiveBench.
Who runs LiveBench Design and how often is it updated?
LiveBench Design is published by LiveBench. BenchLeader re-reads the published results once a day and records the date each result was published.
Does LiveBench Design count toward the BenchLeader Index?
No. LiveBench Design is shown for reference but left out of the composite index.