Read the blog
Manipulation research and benchmarks.
FAQ
Answers about GripSim
The manipulation layer, the products, pricing, and how to get started.
Built on the NVIDIA Physical AI stack
NVIDIA InceptionIsaac LabGR00TJetson ThorOmniverseCosmos
Basics
About GripSim
What is GripSim?
The manipulation layer for robots — a cross-embodiment model, world model, runtime, and agent that give any hand dexterity.
Do you build robots?
No. We give the robots you build or operate human-level grasping.
What's cross-embodiment?
One policy that runs across many gripper and hand shapes.
Products
About the stack
What are the five products?
Dextra (policy), Tactyl (contact world model), Reflex (edge runtime), Simdex (data engine), Sequa (agentic skills).
Can I use one product alone?
Yes — each works standalone and integrates with the rest.
Which hands are supported?
Many morphologies, with new ones added via transfer.
Technology
Under the hood
Why are GPUs essential?
Training uses NVIDIA DGX/HGX; edge inference uses Jetson — both are GPU-bound.
How do you train without real data?
Simdex generates contact-rich data in Isaac Sim; the flywheel adds real data per deployment.
Does it run on-robot?
Yes, Reflex runs the policy at control-loop latency on Jetson.
Pricing & pilots
Getting started
Is there a free tier?
Yes — the Simdex sandbox with a demo policy.
How is production priced?
Per robot/fleet for the runtime, plus usage-based skill API.
How long is a pilot?
Scoped to hit a benchmark on one cell.
Still curious?
More ways to learn
Explore deeper resources.
Browse resources
Guides, docs, and benchmarks.
Contact us
Ask our engineers directly.
5
Products
1
Manipulation brain
4.8
Avg rating
Helpful?
Readers say
“Answered everything I had.”
“Clear and concise.”
“Sent it to my team.”
Can't find it?
Ask us anything
If your question isn't here, our team will answer fast.
- Technical questions
- Pilot scoping
- Partnerships
Why GripSim
Built for production robotics
The manipulation layer teams ship with, not a research demo.
Cross-embodiment by design
One policy runs across many grippers and hands — no per-robot rewrite.
Closed-loop with touch
Tactyl predicts slip and force so grasps react before they fail.
Runs on the robot
Reflex executes on Jetson at control-loop latency, with safety limits.
A moat that compounds
Every deployment feeds the shared model through the data flywheel.
Stay in the loop
Manipulation research, in your inbox
Benchmarks, sim-to-real notes, and product updates. No hype, no spam.