One API. 253 capabilities.
Simulate physics, optimise routes, forecast weather, generate worlds, move robots and read the scene. Call each one by REST, OpenAPI, MCP tools, the command line or Python, and it runs on the Ipnops GPU fleet with no hardware on your side.
All capabilities are called at https://compute.ipnops.ai (REST, OpenAPI at /openapi.json, MCP at /mcp), with an Ipnops API key.
Every capability has a typed request and a typed result. The OpenAPI document lists exactly the capabilities your key can use.
Each capability is a tool an agent can call, so assistants and agent frameworks run jobs directly.
Submit, follow and download from a script or a notebook in a few lines.
A key carries the capabilities it can run, a limit on job time and queue, monthly GPU and CPU hours, and an expiry. A key cannot make a wider key.
Jobs queue and run without a screen. Follow progress as a live event stream and fetch results by signed link.
Robot middleware, industrial servers, motion and vehicle buses, brokers and open file formats connect Ipnops to the machines and systems you already run.
Rigid bodies, cloth, rooms and acoustics, buildings and comfort, crowds, and fast stand-ins for slow physics.
Vehicle routing, graph analysis and quantum-circuit search for the best plan, returned as routes and costs.
Station and gridded forecasts with prediction intervals, verified against what happened.
Scenes rendered, reconstructed and generated from a description or a capture.
Humanoid and robot motion: text to motion, in-betweening, retargeting and policy export.
Detection, tracking, body pose, matting and image to 3D.
Radio coverage maps for sites and cities.
GPU quantum simulation and optimisation circuits.
Protein structure, docking and molecule generation.
Rendered and generated training data with ground truth.
Driving scenarios, paths ahead and front-camera clips.
Get your API key.
Tell us what you want to run. We answer every request ourselves.