List your company

Long-horizon Task Execution on Minecraft Long-horizon Tasks

100WoodHuman
18.8839.946182.06Feb 27, 2025Mar 14, 2025Mar 29, 2025Apr 14, 2025Apr 29, 2025May 14, 2025May 30, 2025

Evaluation results

MethodLinks
Human2025.021001008617163328—
Optimus-2Policy=GOAP2025.029993539132821—
Optimus-2Dependency=Oracle, Planner LLM=GPT-4V2025.05999353913282145
Optimus-1Policy=STEVE-12025.029892468112519—
Optimus-1Dependency=Oracle, Planner LLM=GPT-4V2025.05989246811251943
XENON *Dependency=Oracle, Planner LLM=Qwen2.5-VL-7B2025.059593837375756179
Jarvis-1Policy=STEVE-12025.02938936781615—
Jarvis-1Dependency=Oracle, Planner LLM=GPT-42025.0593893678161538
Optimus-1†Dependency=Oracle, Planner LLM=Qwen2.5-VL-7B2025.0592802291041734
XENONDependency=Learned, Planner LLM=Qwen2.5-VL-7B2025.05858146746402854
DEPSPolicy=STEVE-12025.0277481601010—
DEPSDependency=None, Planner LLM=Codex2025.057748160101022
Pure GPT-4VPolicy=GOAP, Context=zero-shot manner to generate executable sub-goals2025.025031122133—
Pure GPT-4VPolicy=STEVE-1, Context=zero-shot manner to generate executable sub-goals2025.02412000000—
Pure GPT-4VPolicy=VPT (text), Context=zero-shot manner to generate executable sub-goals2025.0222800000—