LiveBench 2.0 Browser Use

Multi-step tasks in working web apps, done through a real browser; share of tasks passed. LiveBench 2.0, refreshed with new questions each release to limit contamination.

CategoryAgents
MetricScore (%)
SourceLiveBench (original)
LicenseNone stated
Benchmark pagelivebench.ai
StatusScored
Weight1.0 within Agents
FitDiscrimination 2.07, noise 0.78

Results

#ModelProviderResultDate
1 GPT-6.1 Sol OpenAI 93.3% 2026-11-06
2 GPT-6.1 Terra OpenAI 93.1% 2026-11-06
3 GPT-6 Astra OpenAI 91.3% 2026-11-06
4 Claude Fable 5.1 Anthropic 89.8% 2026-11-06
5 Claude Sonnet 5.5 Anthropic 89.3% 2026-11-06
6 Claude Opus 5.5 Anthropic 88.4% 2026-11-06
7 Claude Fable 5.2 Anthropic 86.9% 2026-11-06
8 Gemini 4 Argon Google 82.8% 2026-11-06
9 MiniMax-M3.1-Flash MiniMax 80.6% 2026-11-06
10 Gemini 4 Carbon Google 80.0% 2026-11-06
11 MiMo-V2.6-Pro Xiaomi 77.3% 2026-11-06
12 Kimi K3.1 Pro Moonshot AI 75.9% 2026-11-06
13 Kimi K3 Moonshot AI 75.2% 2026-11-06
14 GPT-5.6 Terra OpenAI 70.3% 2026-11-06
15 GPT-6 Luna OpenAI 70.3% 2026-11-06
16 Claude Haiku 5.5 Anthropic 69.5% 2026-11-06
17 DeepSeek V4.1 Flash 1017 DeepSeek 67.2% 2026-11-06
18 Muse Wonder Meta 60.0% 2026-11-06
19 GLM-5.3-Flash Zhipu AI 59.8% 2026-11-06
20 Grok 4.6 xAI 49.4% 2026-11-06
21 Qwen3.8 Max Alibaba 49.4% 2026-11-06
22 Grok 4.7 xAI 49.3% 2026-11-06
23 Gemini 3.8 Flash Google 45.1% 2026-11-06
24 Muse Spark 1.3 Meta 30.9% 2026-11-06
25 Muse Spark Open Meta 26.2% 2026-11-06
26 Step 5 Preview StepFun 24.8% 2026-11-06
27 Qwen3.8 27B Alibaba 24.2% 2026-11-06
28 Nemotron 3.5 Super VL 120B A12B NVIDIA 21.6% 2026-11-06
29 Mistral Large 4 Preview Mistral 17.4% 2026-11-06
30 Inkling Thinking Machines 12.0% 2026-11-06
31 Mistral Large 4 Mistral 0.6% 2026-11-06
32 Gemma 4 31B IT Google 0.0% 2026-11-06