New: Agent Index for knowledge work is live.Explore the index ↗

Best LLMs for long-running agentic work

LiveBench: Agentic

LiveBench 2.0 evaluates long-running agentic work on its October 25, 2026 release. This suite contains 18 tasks or questions, scored from 0 to 100 in one run per model. Models use their highest supported effort unless a different setting is reported. This is an individual task-suite score, separate from LiveBench’s weighted overall average.

Updated: Nov 5, 2026Data as of: Nov 2, 2026

Widely usedNot saturated

Official benchmark

LiveBench: Agentic results. Each row identifies its benchmark version and evaluation track. Scores from different versions are not directly comparable. Best effort from the latest stored report date per model and version.
Model / creator	Version	Reported score
Claude Fable 5.1Anthropic · Proprietary	2.0 / 2026-10-25	77.9%
GPT-6 AstraOpenAI · Proprietary	2.0 / 2026-10-25	77.5%
Claude Opus 5.5Anthropic · Proprietary	2.0 / 2026-10-25	72.8%
GPT-6.1 SolOpenAI · Proprietary	2.0 / 2026-10-25	68.7%
Claude Sonnet 5.5Anthropic · Proprietary	2.0 / 2026-10-25	63.4%
Gemini 4 ArgonGoogle · Proprietary	2.0 / 2026-10-25	61.2%
Muse Wonder 1.0Meta · Proprietary	2.0 / 2026-10-25	52.6%
GPT-6.1 TerraOpenAI · Proprietary	2.0 / 2026-10-25	48.6%
Grok 4.7xAI · Proprietary	2.0 / 2026-10-25	46.1%
Claude Haiku 5.5Anthropic · Proprietary	2.0 / 2026-10-25	40%
Kimi K3.1 ProMoonshot AI · Open source / open weights	2.0 / 2026-10-25	39.1%
GLM-5.3Z.ai · Open source / open weights	2.0 / 2026-10-14	35.9%
DeepSeek V4.1 Flash 1017DeepSeek · Open source / open weights	2.0 / 2026-10-25	35.7%
MiMo-V2.6-ProXiaomi · Open source / open weights	2.0 / 2026-10-25	35.2%
GPT-5.6 TerraOpenAI · Proprietary	2.0 / 2026-10-25	33.1%
GPT-6 LunaOpenAI · Proprietary	2.0 / 2026-10-25	30.9%
MiniMax M3.1 FlashMiniMax · Open source / open weights	2.0 / 2026-10-25	28.8%
Kimi K3Moonshot AI · Open source / open weights	2.0 / 2026-10-25	26.4%
Muse Spark 1.3Meta · Proprietary	2.0 / 2026-10-25	25.5%
Qwen3.8 27BQwen · Open source / open weights	2.0 / 2026-10-25	20.6%
Step 5 PreviewStepFun · Open source / open weights	2.0 / 2026-10-25	19.4%
GLM-5.3-FlashZ.ai · Open source / open weights	2.0 / 2026-10-25	19%
Gemini 3.8 FlashGoogle · Proprietary	2.0 / 2026-10-25	16.5%
Muse Spark OpenMeta · Open source / open weights	2.0 / 2026-10-25	11.7%
Nemotron 3.5 Super VLNVIDIA · Open source / open weights	2.0 / 2026-10-25	11.7%
InklingThinking Machines · Open source / open weights	2.0 / 2026-10-25	8.5%

View all 26 published reports

Muse Wonder 1.052.6%

2.0 / 2026-10-25

release: 2026-10-25 · Highest supported effort unless publisher reports an exception; single run · suite: agentic

GPT-6.1 Terra48.6%

2.0 / 2026-10-25

release: 2026-10-25 · Highest supported effort unless publisher reports an exception; single run · suite: agentic

Claude Haiku 5.540%

2.0 / 2026-10-25

release: 2026-10-25 · Highest supported effort unless publisher reports an exception; single run · suite: agentic

MiMo-V2.6-Pro35.2%

2.0 / 2026-10-25

release: 2026-10-25 · Highest supported effort unless publisher reports an exception; single run · suite: agentic

GPT-5.6 Terra33.1%

2.0 / 2026-10-25

release: 2026-10-25 · Highest supported effort unless publisher reports an exception; single run · suite: agentic

GPT-6 Luna30.9%

2.0 / 2026-10-25

release: 2026-10-25 · Highest supported effort unless publisher reports an exception; single run · suite: agentic

Qwen3.8 27B20.6%

2.0 / 2026-10-25

release: 2026-10-25 · Highest supported effort unless publisher reports an exception; single run · suite: agentic

Step 5 Preview19.4%

2.0 / 2026-10-25

release: 2026-10-25 · Highest supported effort unless publisher reports an exception; single run · suite: agentic

Nemotron 3.5 Super VL11.7%

2.0 / 2026-10-25

release: 2026-10-25 · Highest supported effort unless publisher reports an exception; single run · suite: agentic

Muse Spark Open11.7%

2.0 / 2026-10-25

release: 2026-10-25 · Highest supported effort unless publisher reports an exception; single run · suite: agentic

2.0 / 2026-10-25

release: 2026-10-25 · Highest supported effort unless publisher reports an exception; single run · suite: agentic

Claude Fable 5.177.9%

2.0 / 2026-10-25

release: 2026-10-25 · Highest supported effort unless publisher reports an exception; single run · suite: agentic

GPT-6 Astra77.5%

2.0 / 2026-10-25

release: 2026-10-25 · Highest supported effort unless publisher reports an exception; single run · suite: agentic

Claude Opus 5.572.8%

2.0 / 2026-10-25

release: 2026-10-25 · Highest supported effort unless publisher reports an exception; single run · suite: agentic

GPT-6.1 Sol68.7%

2.0 / 2026-10-25

release: 2026-10-25 · Highest supported effort unless publisher reports an exception; single run · suite: agentic

Claude Sonnet 5.563.4%

2.0 / 2026-10-25

release: 2026-10-25 · Highest supported effort unless publisher reports an exception; single run · suite: agentic

Gemini 4 Argon61.2%

2.0 / 2026-10-25

release: 2026-10-25 · Highest supported effort unless publisher reports an exception; single run · suite: agentic

2.0 / 2026-10-25

release: 2026-10-25 · Highest supported effort unless publisher reports an exception; single run · suite: agentic

Kimi K3.1 Pro39.1%

2.0 / 2026-10-25

release: 2026-10-25 · Highest supported effort unless publisher reports an exception; single run · suite: agentic

DeepSeek V4.1 Flash 101735.7%

2.0 / 2026-10-25

release: 2026-10-25 · Highest supported effort unless publisher reports an exception; single run · suite: agentic

MiniMax M3.1 Flash28.8%

2.0 / 2026-10-25

release: 2026-10-25 · Highest supported effort unless publisher reports an exception; single run · suite: agentic

2.0 / 2026-10-25

release: 2026-10-25 · Highest supported effort unless publisher reports an exception; single run · suite: agentic

Muse Spark 1.325.5%

2.0 / 2026-10-25

release: 2026-10-25 · Highest supported effort unless publisher reports an exception; single run · suite: agentic

GLM-5.3-Flash19%

2.0 / 2026-10-25

release: 2026-10-25 · Highest supported effort unless publisher reports an exception; single run · suite: agentic

Gemini 3.8 Flash16.5%

2.0 / 2026-10-25

release: 2026-10-25 · Highest supported effort unless publisher reports an exception; single run · suite: agentic

2.0 / 2026-10-14

release: 2026-10-14 · Official LiveBench 2.0 row GLM-5.3. One run per suite; effort not explicitly labeled on this row. Historical release, not the October 25 index track. · source model id: glm-5.3 · suite: agentic