PokeBot Releases RoboAPI: Embodied Models Now 'Plug-and-Play'
刚刚，破壳机器人发布RoboAPI！具身模型也能「开箱即用」了
PokeBot, a newly established robotics company, has released its first model product, RoboAPI, which enables zero-shot generalization for embodied AI. The API allows developers to use a simple prompt to direct a robotic arm to perform tasks in unfamiliar environments without prior training. The model, Poke-Aurora, demonstrates strong performance on benchmark tasks, with success rates exceeding those of several competing models.
PokeBot's RoboAPI represents a significant step toward making embodied AI accessible by offering a zero-shot generalization API for physical robots, reducing the need for extensive training and customization. The model's performance on standardized benchmarks suggests it outperforms several existing alternatives, though real-world deployment readiness remains to be tested.
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
86%
View original evidence
在这套评测中，Poke-Aurora 在三类任务上的成功率分别为 86.0%、54.1% 和 87.7%，均高于参与对比的 GPT6 Astra 和两款开源具身模型；两款开源具身模型在三类任务上的成功率都不到 35%。
Open source S11Reported success rate
54.1%
View original evidence
在这套评测中，Poke-Aurora 在三类任务上的成功率分别为 86.0%、54.1% 和 87.7%，均高于参与对比的 GPT6 Astra 和两款开源具身模型；两款开源具身模型在三类任务上的成功率都不到 35%。
Open source S11Reported success rate
87.7%
View original evidence
在这套评测中，Poke-Aurora 在三类任务上的成功率分别为 86.0%、54.1% 和 87.7%，均高于参与对比的 GPT6 Astra 和两款开源具身模型；两款开源具身模型在三类任务上的成功率都不到 35%。
Open source S11Reported success rate
<35%
View original evidence
在这套评测中，Poke-Aurora 在三类任务上的成功率分别为 86.0%、54.1% 和 87.7%，均高于参与对比的 GPT6 Astra 和两款开源具身模型；两款开源具身模型在三类任务上的成功率都不到 35%。
Open source S11
- paper: 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.
在这套评测中，Poke-Aurora 在三类任务上的成功率分别为 86.0%、54.1% 和 87.7%，均高于参与对比的 GPT6 Astra 和两款开源具身模型；两款开源具身模型在三类任务上的成功率都不到 35%。
Open source S11
Source：Machine Heart — Robotics on WeChat · mp.weixin.qq.com