News
Building, shipping, learning.
Product updates, announcements, technical deep dives, event recaps, and perspectives from the PrismaX team.
Deep Dive
Quick Note: π0.5 Inference in 16 GB
Two fixes cut our π0.5 rollout stack from about 22 GB to 11.2 GB of VRAM, so it now fits 16 GB cards like the RTX 5080 and 5070 Ti.
Deep Dive
Latency Versus π0.5 Rollout Performance
A Python pipe buffer added hundreds of milliseconds to our π0.5 rollouts. How we found it, cut inference from 230 ms to 87 ms, and what changed on the robot.
Deep Dive
openpi / π0.5 Fine-Tuning: How-To and Tips
Ablations from a π0.5 fine-tune on YAM pick-and-place data: training scheme, lighting, inference bugs, camera placement and field of view, and backgrounds.
All news
Recap
IROS 2026 Recap
Field notes from IROS 2026 in Pittsburgh: a show floor full of dexterous hands, hand makers turning into data companies, and site-unseen demos that held up.
Announcement
Introducing PrismaX Robotics Data
Usable, diverse, proven demonstrations for the teams building physical AI. Now in limited release.
Perspective
No Robot Is Safe
AI safety started as a commercial problem for LLMs. Robots inherit it, plus physics. We map each LLM safety failure onto physical AI.
Product Update
PrismaX Product Updates: Q2 2026
What shipped in Q2 2026: VLA data uploads for operators, Verify Quality opening to members, and a redesigned Robot Fleet built as a hardware marketplace.
Product Update
Introducing The First 100
Verify Quality opens to the public. Score the robot training data models learn from, build a track record, and compete to become one of the founding Data Quality Validators on PrismaX.
Recap
A year of robotics in 10 minutes
Hardware grew up, video-only training went mainstream, and closed-loop revenue is still missing. A quick read on what changed in robotics this year and what we want to see next.
Recap
ICRA 2026 Recap
Field notes from ICRA 2026 in Vienna: no new hardware by design, money moving toward commercial deployment, and the data layer that makes robots worth anything once they ship.
Perspective
Standards for Data
The first arc of the PrismaX Standards for Data: repeatable guidelines for collecting pretraining-grade robotics data, and the reference our validators use to apply them.
Perspective
Not All Robotics Data Is Created Equal
A primer on the two paths of physical AI, the data each one actually needs, and what data quality looks like at the technical level.
Product Update
Introducing the PrismaX Regional Ambassador Program
PrismaX is opening applications for its first cohort of Regional Ambassadors.
Product Update
PrismaX Product Updates: Q1 2026
A look at what shipped in Q1 2026, from faster and more reliable sessions to new arms, improved queue visibility, and updates to the Robot Control Center.
Announcement
PrismaX Joins NVIDIA Inception Program
PrismaX Joins NVIDIA Inception Program to Accelerate the Service Layer for Physical AI.
Perspective
The Service Layer Era
PrismaX is the service layer for physical AI, setting standards for how robots, data, and human work are deployed to advance intelligence.
Deep Dive
Intro to AI for Robotics
An introduction to AI-native robotics, covering learning loops, teleoperation data, and how modern robots move from demos to real work.
Deep Dive
Teleoperation: Now and the Future of AI Robotics
Read this article if you’re interested in contributing to the robotics revolution, which will reshape the way we work and interact with the physical world.
Announcement
PrismaX Launches AI Teleoperations Platform
PrismaX launches its AI teleoperations platform, enabling remote robot control and paving the way for mainstream robotics adoption.
Perspective
PrismaX: The Base Layer for AI Robotics
PrismaX builds the base layer for real-world AI robotics, uniting data, teleoperation, and models to accelerate autonomous robots.
Announcement
PrismaX Raises $11M: a16z CSX AI Robotics Funding
PrismaX emerges with $11M led by a16z CSX to build decentralized AI robotics, creating fair-use data standards and scalable teleoperation.
Deep Dive
GR00T Part 4: Advancing Robotics Foundation Models
In this final part, we explore how robotics foundation models can move from GPT-2 to GPT-3 era using synthetic, simulator, and high-quality data.
Deep Dive
GR00T Part 3: Predictive Control in AI Robotics
Exploring how predictive models and latent actions help robots learn tasks—from walking to making sandwiches—paving the way for scalable AI robotics.
Deep Dive
GR00T Part 2: Labels Without Labels
Unsupervised training teaches AI to extract powerful features from video, linking compression to real-world actions and advancing robotics learning.
Deep Dive
GR00T Part 1: Foundation Models & the History of AI Robotics
Nvidia’s GR00T-1 marks a GPT2 moment for robotics—proof that foundation-scale models trained on video can generalize across tasks and hardware.
Robotics Data
The data behind the deep dives.
Every experiment we write up trains on the same QA-scored data we deliver to teams.