LiveBench 2.0 Design
Professional Work · 2026-10-14
LiveBench 2.0 Design suite: Produce or rework a finished document to a brief: Word documents, PDFs and slide decks; 25 tasks. LiveBench 2.0 (Abacus AI, released October 14, 2026) replaced the v1 question-answer categories with seven suites, five of them agentic; each suite is one run per model at its highest supported effort, a task the run could not measure scores 0, and a model without an input a suite needs (a text-only model on Browser) scores 0 there. Not comparable with LiveBench v1.
Top models (higher is better)
| Model | Score |
|---|---|
| Claude Opus 5.5 | 86.9 |
| Claude Sonnet 5.5 | 85.0 |
| Claude Fable 5 | 84.7 |
| Claude Fable 5.1 | 83.4 |
| DeepSeek V4.1 Flash 1017 | 83.4 |
| Claude Haiku 5.5 | 83.1 |
| Claude Opus 5 | 83.0 |
| GPT-6 Luna | 82.5 |
| GPT-6 Sol | 82.2 |
| GPT-6.1 Sol | 82.1 |
| Sonnet 5 | 82.0 |
| GPT-6 Astra | 82.0 |
| GPT-5.6 Sol | 81.6 |
| Grok 4.6 | 79.8 |
| DeepSeek V4.1 Flash | 79.2 |
| Opus 4.8 | 78.8 |
| MiMo-V2.6-Pro | 78.5 |
| Grok 4.5 | 78.1 |
| Grok 4.7 | 78.0 |
| Kimi K3.1 Pro | 77.8 |
| GLM-5.3 | 77.3 |
| GLM-5.3 Flash | 77.2 |
| Kimi K3 | 77.2 |
| Step 5 Preview | 76.9 |
| GPT-5.6 Luna | 76.0 |
| GPT-5.6 Terra | 76.0 |
| Qwen3.8 27B | 75.9 |
| Muse Wonder | 75.1 |
| Muse Spark 1.3 | 74.8 |
| MiniMax M3.1 Flash | 74.0 |
| Gemini 4 Argon | 73.7 |
| Gemini 3.7 Flash | 73.2 |
| Qwen3.8-Flash-Next | 72.8 |
| Gemini 3.8 Flash | 72.7 |
| Sonnet 4.6 | 71.8 |
| DeepSeek-V4-Flash-0731 | 71.7 |
| DeepSeek V4 Flash Vision Exp | 70.7 |
| GPT-5.5 | 70.3 |
| Beam 1.0 501B A23B | 67.9 |
| Muse Spark 1.2 | 67.1 |