Battle at RoboColiseum. GPT, Claude, Gemini Robot Capabilities Clash, Surprises Await After Close Wins
对决RoboColiseum！GPT、Claude、Gemini机器人能力PK，险胜之后还有黑马
RoboColiseum releases an evaluation report, comparing the performance of multimodal models such as GPT, Claude, and Gemini with specialized embodied strategy models (such as Paimon, HiDream-O1-Embodied) in dimensions such as instruction following, spatial understanding, perturbation adaptation, and general operations.
Source: Machine Heart — Robotics on WeChat · Read original article ↗
English translation unavailable. Choose the original article to read it.
What the source reports
Publisher-reported claims, with original evidence. These results have not been independently verified by RoboSignal.
Reported numbers
Reported success rate
90%
Reported success rate
80%
average planning decisions
16.7
average planning decisions
35.6
average action steps
118.8
average action steps
158.2
Reported success rate
<10%
View original evidence
平台公布的验证数据显示，在已验证的模型和任务范围内，仿真评测与实机部署表现的相关性达到 89.5%，相同模型在仿真与真机任务中的评测成功率差异小于 10%。
Open source S22
- paper: not_reported
- code: not_reported
- weights: not_reported
- dataset: not_reported
What remains unknown
Not established in the collected evidence: Environment, Control, Data origin.
Reported performance applies to the described task. It does not establish general autonomy or deployment readiness.
Source excerpts and review record
Automatically extracted; no manual editorial approval recorded.
完整配置下，10 个任务的成功率为 90%，平均规划决策 16.7 次。
Open source S12
移除明确的执行偏差反馈后，成功率降至 80%，平均规划决策增至 35.6 次。
Open source S13
移除深度输入和查询工具后，任务成功率同样从 90% 降至 80%，平均下发动作步数从 118.8 步增至 158.2 步。
Open source S15
平台公布的验证数据显示，在已验证的模型和任务范围内，仿真评测与实机部署表现的相关性达到 89.5%，相同模型在仿真与真机任务中的评测成功率差异小于 10%。
Open source S22
Source：Machine Heart — Robotics on WeChat · mp.weixin.qq.com