Primotion, a Shenzhen-based startup, has introduced its embodied AI foundation model, Paimon-0. This model reportedly achieves action-chunk inference speeds of 300 milliseconds, a tenfold improvement according to the company. Primotion aims to build robot chips by first developing its own AI models, with Paimon-0 already leading several embodied AI benchmarks. AI
IMPACT This model's rapid inference could accelerate the development and deployment of more responsive and capable robots.
RANK_REASON New embodied AI foundation model release from a startup. [lever_c_demoted from frontier_release: ic=1 ai=1.0]
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →