LiveBench 2.0 Agentic

Long-running work: business apps extended over several turns of changing requirements, codebase operations over simulated months, extended coding and optimisation tasks; deterministic grading. LiveBench 2.0, refreshed with new questions each release to limit contamination.

CategoryAgents
MetricScore (%)
SourceLiveBench (original)
LicenseNone stated
Benchmark pagelivebench.ai
StatusScored
Weight1.1 within Agents
FitDiscrimination 1.21, noise 0.46

Results

#ModelProviderResultDate
1 Claude Fable 5.2 Anthropic 82.9% 2026-11-06
2 GPT-6 Astra OpenAI 75.6% 2026-11-06
3 Claude Fable 5.1 Anthropic 74.2% 2026-11-06
4 Claude Opus 5.5 Anthropic 72.8% 2026-11-06
5 Gemini 4 Carbon Google 72.7% 2026-11-06
6 GPT-6.1 Sol OpenAI 65.6% 2026-11-06
7 Claude Sonnet 5.5 Anthropic 63.4% 2026-11-06
8 Gemini 4 Argon Google 58.5% 2026-11-06
9 GPT-6.1 Terra OpenAI 56.2% 2026-11-06
10 Muse Spark 1.3 Meta 47.9% 2026-11-06
11 Claude Haiku 5.5 Anthropic 47.9% 2026-11-06
12 Muse Wonder Meta 46.2% 2026-11-06
13 Grok 4.7 xAI 45.3% 2026-11-06
14 DeepSeek V4.1 Flash 1017 DeepSeek 38.8% 2026-11-06
15 Kimi K3.1 Pro Moonshot AI 38.7% 2026-11-06
16 MiMo-V2.6-Pro Xiaomi 36.4% 2026-11-06
17 Kimi K3 Moonshot AI 36.0% 2026-11-06
18 Qwen3.8 Max Alibaba 26.6% 2026-11-06
19 GPT-5.6 Terra OpenAI 25.9% 2026-11-06
20 MiniMax-M3.1-Flash MiniMax 25.7% 2026-11-06
21 GPT-6 Luna OpenAI 24.7% 2026-11-06
22 Mistral Large 4 Preview Mistral 24.7% 2026-11-06
23 GLM-5.3-Flash Zhipu AI 20.7% 2026-11-06
24 Step 5 Preview StepFun 19.4% 2026-11-06
25 Grok 4.6 xAI 18.7% 2026-11-06
26 Gemini 3.8 Flash Google 16.5% 2026-11-06
27 Qwen3.8 27B Alibaba 15.9% 2026-11-06
28 Muse Spark Open Meta 12.1% 2026-11-06
29 Nemotron 3.5 Super VL 120B A12B NVIDIA 12.0% 2026-11-06
30 Inkling Thinking Machines 9.8% 2026-11-06
31 Mistral Large 4 Mistral 9.2% 2026-11-06
32 Gemma 4 31B IT Google 8.5% 2026-11-06