OpenBMB says MiniCPM-Robot jumps from 10 Hz to 36 Hz with CUDA Graphs
iamfakhrealam · x · 2026-07-24
OpenBMB says PhyAI's CUDA Graph optimization and custom Triton fused kernels lift MiniCPM-Robot inference from 10 Hz to 33 Hz, and to 36 Hz on an NVIDIA H20.
The post argues that the key issue in embodied AI is not just perception, but making multi-step, context-aware inference cheap enough to run at robot speed. It also emphasizes that the release is fully open source.
Related event: OpenBMB Open-Sources MiniCPM-Robot for Offline Embodied AI(10 posts)→
More from Embodied
- NVIDIA open-sources FLORA, a robotics data engine ~125x cheaper than GPT-6 Astra — nvidia · 2026-10-27
- Embodied AI startups need a different operational discipline, investor says — ___Mufasaa · 2026-10-27
- π0.5 study: raising demos from 10 to 438 lifts Place Ball in Bowl success from 7% to 83% — stingning · 2026-10-27
- GPT-6 Astra controls dexterous hands zero-shot at just 5.2% success in 500 trials — stingning · 2026-10-27
- HandsOn benchmark: 66 verified tasks and 18,000 human demos for dexterous manipulation — stingning · 2026-10-27
- DexPort: a handheld interface with a real robot hand for dexterous-hand data collection — pabbeel · 2026-10-27