Native Robot Reasoning: leaderboard
Vision-language models control robots on modified LIBERO tasks and real-world laboratory tasks. Scores separate task success from partial progress and retain each model’s reasoning level. Published general-capability checks of the released base model are separate from its trained robot policy.
Source: arxiv.org.
Benchmark results
Results are published separately for each benchmark or evaluation setting.
- Native Robot Reasoning - General Baseline - AIME 2026
- Native Robot Reasoning - General Baseline - ERQA
- Native Robot Reasoning - General Baseline - GPQA Diamond
- Native Robot Reasoning - General Baseline - IFBench
- Native Robot Reasoning - General Baseline - LiveCodeBench v6
- Native Robot Reasoning - General Baseline - MMMU (val)
- Native Robot Reasoning - General Baseline - MMMU-Pro standard
- Native Robot Reasoning - General Baseline - MMMU-Pro vision
- Native Robot Reasoning - General Baseline - RealWorldQA
- Native Robot Reasoning - LIBERO - Average
- Native Robot Reasoning - LIBERO - Goal
- Native Robot Reasoning - LIBERO - Long
- Native Robot Reasoning - LIBERO - Object
- Native Robot Reasoning - LIBERO - Spatial
- Native Robot Reasoning - RoboLab-120 - All
- Native Robot Reasoning - RoboLab-120 - Complex
- Native Robot Reasoning - RoboLab-120 - Moderate
- Native Robot Reasoning - RoboLab-120 - Progress Score
- Native Robot Reasoning - RoboLab-120 - Simple
Interactive version: theaggregate.ai/benchmark?slug=native-robot-reasoning · How It Works · Data refreshed daily, snapshot 2026-11-17.