LiveBench design leaderboard
- Claude Opus 5.5 leads
- 86.9%
- models measured
- 43
- lead over #2
- +1.9 pts
- median
- 77.2%
Leader on LiveBench design · read October 21
On October 21, 2026, 43 of our catalog's models had a LiveBench design result, and Anthropic Claude Opus 5.5 led with 86.9%, ahead of Anthropic Claude Sonnet 5.5 at 85.0%.
Layout and design tasks across presentations, PDFs and documents, graded on visual and structural checks of the produced file. Each score is the model's mean task score for the category, 0 to 100, from a single run of the current release, shown as a share.
Role in the index
LiveBench design is shown as published and does not feed any of our indexes. Read the methodology
| # | Model | Result | Cost per task | Output tokens per task | Evidence | Coding index | Output $/Mtok |
|---|---|---|---|---|---|---|---|
| 1 | Claude Opus 5.5Anthropic | 86.9% | $5.7483 | 150,005 | 88.8 | $20.00 | |
| 2 | Claude Sonnet 5.5Anthropic | 85.0% | $5.0166 | 210,867 | 87.3 | $10.00 | |
| 3 | Claude Fable 5Anthropic | 84.7% | $8.5495 | 63,307 | 79.3 | $50.00 | |
| 4 | Claude Fable 5.1Anthropic | 83.4% | $7.3755 | 59,570 | 84.2 | $50.00 | |
| 5 | DeepSeek V4.1 Flash 1017DeepSeek | 83.4% | $0.212 | 113,066 | 70.3 | $0.60 | |
| 6 | Claude Haiku 5.5Anthropic | 83.1% | $0.6631 | 171,437 | 70.7 | $0.50 | |
| 7 | Claude Opus 5Anthropic | 83.0% | $5.4526 | 58,259 | 83.0 | $25.00 | |
| 8 | GPT-6 LunaOpenAI | 82.5% | $0.0306 | 44,762 | 57.3 | $0.50 | |
| 9 | GPT-6 SolOpenAI | 82.2% | $0.4254 | 28,387 | 76.0 | $10.00 | |
| 10 | GPT-6.1 SolOpenAI | 82.1% | $0.3705 | 25,521 | 82.4 | $10.00 | |
| 11 | Claude Sonnet 5Anthropic | 82.0% | $5.4721 | 108,233 | 57.7 | $10.00 | |
| 12 | GPT-6 AstraOpenAI | 82.0% | $2.8756 | 24,627 | 85.1 | $50.00 | |
| 13 | GPT-5.6 SolOpenAI | 81.6% | $1.7062 | 25,152 | 73.5 | $20.00 | |
| 14 | Grok 4.6xAI | 79.8% | $1.2666 | 46,787 | 65.7 | $6.00 | |
| 15 | DeepSeek V4.1 FlashDeepSeek | 79.2% | $0.1748 | 90,915 | 62.5 | $0.78 | |
| 16 | Claude Opus 4.8Anthropic | 78.8% | $6.4503 | 81,857 | 65.0 | $25.00 | |
| 17 | Qwen3.8 Max (0902)Qwen | 78.6% | $0.5962 | 45,551 | 62.4 | $6.00 | |
| 18 | MiMo-V2.6-ProXiaomi | 78.5% | $0.0932 | 58,295 | 65.4 | $0.87 | |
| 19 | Grok 4.5xAI | 78.1% | $0.3877 | 24,466 | 58.6 | $6.00 | |
| 20 | Grok 4.7xAI | 78.0% | $3.3696 | 117,654 | 66.7 | $6.00 | |
| 21 | GLM 5.3Z.ai | 77.3% | $0.9556 | 68,015 | 70.6 | $4.40 | |
| 22 | Kimi K3Moonshot AI | 77.2% | $1.0007 | 31,713 | 62.3 | $15.00 | |
| 23 | GLM 5.3 FlashZ.ai | 77.2% | $0.0634 | 38,912 | 57.7 | $0.50 | |
| 24 | Step 5 PreviewStepFun | 76.9% | $0.7457 | 71,830 | 61.1 | $2.70 | |
| 25 | GPT-5.6 LunaOpenAI | 76.0% | $0.0895 | 27,152 | 58.0 | $1.20 | |
| 26 | GPT-5.6 TerraOpenAI | 76.0% | $0.6201 | 23,164 | 63.9 | $12.00 | |
| 27 | Qwen3.8 27BQwen | 75.9% | $0.3836 | 63,352 | 74.2 | $2.55 | |
| 28 | Muse Spark 1.3Meta | 74.8% | $0.8286 | 48,559 | 57.4 | $4.25 | |
| 29 | Gemini 4 ArgonGoogle | 73.7% | $1.1604 | 60,890 | 81.0 | $10.00 | |
| 30 | Gemini 3.7 FlashGoogle | 73.2% | $0.6231 | 62,179 | 58.0 | $3.75 | |
| 31 | Qwen3.8 FlashQwen | 72.8% | $0.1281 | 84,253 | Not benchmarked | $0.47 | |
| 32 | Gemini 3.8 FlashGoogle | 72.7% | $0.9796 | 86,732 | 61.6 | $3.75 | |
| 33 | Claude Sonnet 4.6Anthropic | 71.8% | $2.1386 | 52,865 | 57.3 | $15.00 | |
| 34 | DeepSeek V4 Flash Vision ExpDeepSeek | 70.7% | $0.0594 | 14,589 | Not benchmarked | $1.32 | |
| 35 | GPT-5.5OpenAI | 70.3% | $1.1904 | 17,832 | 71.3 | $30.00 | |
| 36 | Muse Spark 1.2Meta | 67.1% | $0.8562 | 44,091 | 55.7 | $4.25 | |
| 37 | MiniMax M3MiniMax | 66.2% | $0.2079 | 46,783 | 39.2 | $1.20 | |
| 38 | Mistral Large 4Mistral AI | 65.6% | $0.389 | 123,574 | 56.4 | $4.18 | |
| 39 | Nemotron 3.5 Super VLNVIDIA | 60.7% | $0.0558 | 32,708 | Not benchmarked | $0.50 | |
| 40 | Muse Spark OpenMeta | 60.3% | $0.3559 | 32,098 | 47.4 | $4.25 | |
| 41 | Claude Haiku 4.5Anthropic | 59.9% | $0.1558 | 11,478 | 35.6 | $5.00 | |
| 42 | Gemma 4 31BGoogle | 58.4% | $0.0148 | 9,942 | 39.0 | $0.40 | |
| 43 | T InklingThinking Machines | 58.3% | $0.2861 | 27,035 | 31.5 | $4.05 |
Only models in our catalog are listed, and each appears once with its best published run. 43 rows were read from the source on October 21, 2026, and rows that match no catalog model are left out. Cost per task is what LiveBench reports the run spent on the category, divided by the number of tasks in it, in US dollars; LiveBench defines it the same way. Every other column is a figure the board publishes for the same run.
| Model | Result | Documents (DOCX) | Presentations (PPT) | |
|---|---|---|---|---|
| Anthropic Claude Opus 5.5 | 86.9% | 90.3% | 88.2% | 84.6% |
| Anthropic Claude Sonnet 5.5 | 85.0% | 94.1% | 87.6% | 79.4% |
| Anthropic Claude Fable 5 | 84.7% | 86.2% | 86.2% | 83.1% |
| Anthropic Claude Fable 5.1 | 83.4% | 92.1% | 84.2% | 79.2% |
| DeepSeek V4.1 Flash 1017 | 83.4% | 81.3% | 89.2% | 80.4% |
| Anthropic Claude Haiku 5.5 | 83.1% | 92.4% | 87.3% | 76.4% |
| Anthropic Claude Opus 5 | 83.0% | 84.1% | 86.4% | 80.3% |
| OpenAI GPT-6 Luna | 82.5% | 86.5% | 78.9% | 83.2% |
| OpenAI GPT-6 Sol | 82.2% | 89.4% | 83.0% | 78.7% |
| OpenAI GPT-6.1 Sol | 82.1% | 86.2% | 79.1% | 82.3% |
| Anthropic Claude Sonnet 5 | 82.0% | 79.2% | 81.5% | 83.5% |
| OpenAI GPT-6 Astra | 82.0% | 84.7% | 78.3% | 83.4% |
| OpenAI GPT-5.6 Sol | 81.6% | 85.2% | 78.0% | 82.4% |
| SpaceXAI Grok 4.6 | 79.8% | 76.9% | 82.0% | 79.6% |
| DeepSeek V4.1 Flash | 79.2% | 74.9% | 86.8% | 75.8% |
| Anthropic Claude Opus 4.8 | 78.8% | 81.0% | 78.6% | 78.1% |
| Qwen3.8 Max (0902) | 78.6% | 78.7% | 80.0% | 77.5% |
| Xiaomi MiMo-V2.6-Pro | 78.5% | 86.5% | 80.0% | 74.1% |
| SpaceXAI Grok 4.5 | 78.1% | 82.4% | 68.8% | 82.6% |
| SpaceXAI Grok 4.7 | 78.0% | 72.9% | 80.1% | 78.6% |
| Z.ai GLM 5.3 | 77.3% | 82.8% | 75.5% | 76.1% |
| MoonshotAI Kimi K3 | 77.2% | 81.4% | 79.0% | 74.2% |
| Z.ai GLM 5.3 Flash | 77.2% | 72.7% | 78.9% | 78.0% |
| StepFun Step 5 Preview | 76.9% | 82.7% | 72.9% | 77.1% |
| OpenAI GPT-5.6 Luna | 76.0% | 77.4% | 78.5% | 73.7% |
| OpenAI GPT-5.6 Terra | 76.0% | 79.2% | 75.5% | 75.0% |
| Qwen3.8 27B | 75.9% | 79.9% | 82.3% | 69.9% |
| Meta Muse Spark 1.3 | 74.8% | 73.4% | 81.1% | 71.1% |
| Google Gemini 4 Argon | 73.7% | 84.5% | 74.6% | 68.6% |
| Google Gemini 3.7 Flash | 73.2% | 80.9% | 76.1% | 68.1% |
| Qwen3.8 Flash | 72.8% | 81.6% | 70.8% | 70.6% |
| Google Gemini 3.8 Flash | 72.7% | 83.1% | 77.8% | 64.9% |
| Anthropic Claude Sonnet 4.6 | 71.8% | 66.8% | 70.0% | 75.1% |
| DeepSeek V4 Flash Vision Exp | 70.7% | 80.0% | 61.8% | 72.7% |
| OpenAI GPT-5.5 | 70.3% | 67.5% | 66.5% | 74.0% |
| Meta Muse Spark 1.2 | 67.1% | 83.9% | 71.5% | 57.1% |
| MiniMax M3 | 66.2% | 73.5% | 65.4% | 63.8% |
| Mistral Large 4 | 65.6% | 81.1% | 52.1% | 68.1% |
| NVIDIA Nemotron 3.5 Super VL | 60.7% | 75.2% | 57.8% | 56.7% |
| Meta Muse Spark Open | 60.3% | 61.3% | 67.1% | 55.3% |
| Anthropic Claude Haiku 4.5 | 59.9% | 70.2% | 54.6% | 59.1% |
| Google Gemma 4 31B | 58.4% | 64.6% | 54.9% | 58.1% |
| Thinking Machines Inkling | 58.3% | 72.5% | 50.5% | 57.5% |
The parts as LiveBench publishes them, for the same run as the result.