MACE-THOR (EmboTeam) - Parallel-Independent Tasks: leaderboard
MACE-THOR (Zeng, Wang and Li, 2026) is a set of long-horizon household tasks for heterogeneous multi-robot teams of two to four robots in the AI2-THOR simulator. Success rate (0-1) on the 21 parallel-independent tasks, where the robots work on separable subtasks; higher is better. The backbone LLM drives the authors' EmboTeam planner, which turns the instruction into a PDDL problem, plans with a classical planner and executes the plan as behavior trees.
Metric: Success rate (0-1; share of the 21 parallel-independent tasks fully completed in AI2-THOR; EmboTeam planner). Source: arxiv.org. Saturation forecast: Estimated already saturated. 3 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-4o | 0.71 | #741 |
| 2 | Claude 3.5 Sonnet | 0.57 | #747 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=mace-thor-emboteam-parallel-independent-tasks · How It Works · Data refreshed daily, snapshot 2026-11-12.