Everything Ipnops runs.
Every capability by domain, each one a single call on the Ipnops GPU fleet.
Re-pose a jointed object from an image into new configurations.
Temperature, daylight and air quality for a room, from its walls and weather.
Convert an IFC building model to mesh and USD.
Check an IFC building model: schema, storeys, elements, issues.
Convert between CAD and mesh formats.
Solids, faces, volume and validity of a part.
Drape a garment on a body and read the fit.
Settle fabric under gravity into its resting shape.
Temperature, maturity and strength gain of a pour from its mix and weather.
Fly a simulated drone and read back its stability and tracking.
Fields and transmission through structures from antennas to optics.
A fast stand-in for porous-flow simulation.
Differentiable simulation examples on GPU.
Concentration downwind of a source for a given wind and stability.
Simulate rigid bodies, soft bodies and fluids in one scene.
Inside temperature and ventilation needed for a target, with evaporative cooling limits.
Temperature spreading through a solid over time.
Precise field solutions on curved geometry with fewer elements.
A temperature column that mixes and freezes, day by day.
Simulate millions of atoms and read back energies and trajectories.
Replace a slow robot simulator with a fast learned model of the same machine.
Simulated laser scans of a scene with ranges and noise.
Solve large linear and mixed-integer planning problems on GPUs and read back the best plan.
Broadcast loudness and true peak to the standard, in one call.
Roll out thousands of physics environments at once for fast policy search.
Relax crystals and molecules; lattice constants, energies and forces.
Where a satellite is, and when it passes over a place.
Run a ready physics scene.
Export a trained policy to ONNX for the machine.
Replay a trained policy and record what it did.
Score a policy over many episodes.
Solve field equations at chosen points without meshing the whole domain.
Simulate falling, colliding and stacking objects for robots and games.
Simulate bodies, contacts and joints at speed.
Train agents among simulated crowds and tasks, scaled on GPUs.
Simulate a robot with its sensors in a world and read back the logs.
Reverberation, clarity and definition per band, ready for a report.
Hear a room before it is built: impulse responses per octave band from geometry and materials.
Rainfall to flows and flooding in pipes and channels of a catchment.
Fit a fast model to your slow simulation.
Answer in milliseconds what the solver takes hours to say.
Run a task end to end to see it work.
List the training tasks you can run.
Contract large networks of tensors for quantum and physics problems on GPUs.
A part becomes a solver-ready mesh at your element size.
Measure training speed on your task.
Wave shoaling, longshore transport, shoreline retreat and a sediment budget for a coast.
Faster language-model serving on the fleet.
Project age groups under survival and birth rates.
Columnar tables for fast analysis of large project data.
Classify behaviour from accelerometer windows.
Oxygen, temperature and waste in a pond or pen.
Credit ledgers with an uncertainty band, a buffer and the price at which a project breaks even.
The effect of an intervention, with confidence intervals and refutation tests, from tables or from before-and-after panels.
Temperature history and remaining shelf life of food.
Schedules and assignments that obey your rules, solved fast.
Biomass and yield of a crop through the season from weather and management.
Infection pressure on a crop from weather and canopy.
Vectors for search and memory, open models on CPU.
Susceptible, infected and recovered over time from the rates you state: peak, final size and a mass check.
Irrigation need, storage and runoff across a farm.
How a fire moves across land from fuel, wind and slope.
Growth of farmed stock from temperature, feed and density.
How infection moves through a farmed or wild fish stock.
How a fish stock grows, is caught and recovers.
Stand volume, thinning, harvest and carbon of a forest.
Components, communities, PageRank and betweenness on GPUs.
Train models that learn from networks of connected things.
Store connected data and answer questions across relationships.
When a crop is ready and what the harvest is worth.
Herd size, growth and feed need over years.
Simulate individual animals or plants and read the group that emerges.
Carbon in pools and harvested wood against a reference level.
Chat completions served on Ipnops GPUs.
Solve linear and integer plans and read back the optimum and its sensitivity.
Growth and output of animals under feed and climate.
Informed traders, noise traders and a market maker, plus auctions, bank networks, supply-chain input-output and growth: price impact, profit, payments and output with the textbook answer beside each run.
Compare a measurement method with a reference: correlation, limits of agreement and repeatability.
Growth curves of microbes under temperature and time.
Fine-tune a language model on your data and score it on a held-out set.
Buckets with per-project access, lifecycle, encryption and quotas.
Where a species can thrive from gridded ocean fields.
Walk crowds through buildings and streets and read flows and bottlenecks.
Put policies against each other in the same tasks and rank them.
Temperature, ventilation and bird comfort inside a poultry house.
Two interacting species through time from the rates you state.
Durable queues and streams between your devices, jobs and services.
Heat stress, bleaching and reef growth from sea temperature.
Vehicle routing with demands and capacities: routes, vehicles and cost.
Screen text before it reaches a model or a person.
Growing degree days, frost, heat stress, chill hours and fire weather.
Organic and inorganic carbon pools in soil and the samples needed to see change.
A soil property map from a few samples, exact at the samples.
Run a standard set of manipulation tasks against a policy.
Storage, operations and egress counted per project and held to its limits.
Store embeddings and search them by meaning, with keyword search alongside.
Fields, water, crops and animals of one farm run together.
Prediction intervals at the coverage you ask for.
Error, bias and anomaly correlation against analysis.
A learned global atmosphere model for fast outlooks.
Learned outlooks for the next ten days across the globe.
A posterior over simulator parameters from runs and one observation.
Predict from a table with no training run.
A second tabular predictor to compare against.
Sensor, meter and log series forecast with intervals.
A second forecaster to compare against.
Coverage, sharpness and CRPS for any forecast, and the checks economists use on prices and incomes: fat tails, volatility clustering, Gini and forecast skill.
Coarse forecasts refined to local detail.
Forecasts from station and gridded data with prediction intervals.
Convert between 3D formats.
Render a frame range to a clip.
Matte a whole capture.
Recover camera paths and depth from a plain video.
Render driving scenes from learned video models for closed-loop tests.
Turn a text or image into a 3D world you can walk through.
Images from a description.
Detailed images from a description for design and research.
Foreground cut-outs with transparency.
Sharpen and improve an image.
A picture becomes a 3D model.
Clean and finish a 3D mesh.
Compare two outputs on the same scorer.
One body model that fits any shape and pose.
Train a splat scene from photos and view it from any angle.
Catch energy from nowhere, bodies below ground and wrong gravity.
Render a person in new lighting from a short capture.
Render a scene to images.
Prims, bounds and validity of a USD scene.
A capture becomes a navigable scene.
Fill rooms and shelves with objects into ready-to-simulate scenes.
Camera and depth renders from a simulated scene.
Scene-by-scene physical plausibility of simulated motion.
Turn a simulator render into realistic footage.
Move scenes between splat formats.
Dense depth from a pair of camera images.
Restyle a clip while the structure holds.
Longer structure-guided clips.
Rendered training data with ground truth.
Generate an animated 3D asset from a sentence.
Video from an image and a prompt.
Turn a clip of a moving object into an animated 3D model.
Worlds and scenes generated from a description.
Generate video of a world from an image, a prompt and your actions.
Score a driving plan for safety and progress.
A short path ahead from one front-camera picture and a speed.
Replay recorded traffic and test a planner against it.
A front-camera clip from a map-style still and a weather prompt.
Drive a rendered town: front-camera frames and trajectories.
Train and test driving policies in generated roads and traffic.
Export a model for the device that runs it.
Estimate, generate and edit human motion from video, text and music.
Candidate grasps for an object from its shape.
Where and how a hand would hold an object from its picture.
Move through simulated homes and interact with objects.
Pretrained humanoid controllers you adapt to your task.
One policy that follows rewards, goals and tracked poses.
Humanoid motion clips ready to use.
A ready policy for a humanoid to walk and balance.
Train and run policies learned from robot demonstrations.
Navigate a simulated building from sight alone.
Generate humanoid walking-and-carrying training data.
Teach a simulated humanoid to copy recorded motion.
Fill the motion between two humanoid poses in real time.
Compose a motion from a timeline of text prompts.
Test driving agents among other interacting vehicles.
A simulated walker that follows paths among people.
Generate pedestrian paths that follow a guide.
Read ROS bags and robot descriptions.
Move demonstrations between robot dataset formats.
Language-driven actions for a robot from images and a task.
Collision-free arm motion to a goal pose.
Fit a robot policy to your demonstrations.
Actions for a humanoid from images and a task.
Consistency check of a robot-learning dataset description: cameras, action sizes, hours and faults.
Physically simulated players for sports training data.
Body motion for a full skeleton from a sentence.
Humanoid motion from a sentence.
Whole-body control of a humanoid from motion commands.
Label every object in a reconstructed scene.
Faster model inference on the fleet.
Create a body from measurements.
Joint coordinates with confidence for every person.
Show what changed between two images of the same place.
Find and follow every person and object in a clip.
Animate a face from audio.
Cut patterns for a garment.
Score 3D hand pose against ground truth.
Head orientation from images without labelled data.
A global 3D body mesh for every person in a video.
Describe and answer questions about images.
Test a policy across varied kitchens.
Higher-quality text vectors for search and memory.
Test a policy as it learns task after task.
Compare an estimated path and map with the true ones.
Position of printed markers in a camera view.
Position and orientation of an object in 3D from one image.
Flag when people and machines come too close.
Replay recorded sensor streams through the pipeline.
Score speech quality.
Text to natural speech.
Speech to text.
Pose of objects the system has never seen, from a model or a few views.
Vectors that describe what happens in a video, for search and prediction.
A second speech voice route.
Signal strength across a scene in moments, with reflections.
Signal strength across a site or city.
Path loss, signal margin and delay for a satellite link from its frequency, altitude and power.
Decode quantum error-correction syndromes with learned and graph decoders.
Decode surface-code errors by minimum-weight matching.
Entangled-state circuits with shot counts on GPU.
Graph-cut search with optimisation circuits.
Contract quantum circuits as tensor networks.
Score and generate DNA sequences.
Ranked poses for a ligand in a protein pocket.
New molecules for a target.
Designed proteins that refold to the target.
Structure from a sequence.
mmCIF structure with per-residue confidence.
Sequence-scale analysis of DNA.
A volume file in, a labelled map and structure volumes out.
Every capability as tools an agent can call, with scoped keys and no passwords.
Every action on a project recorded with who and when.
The same capabilities from a command line and client libraries.
Deploy a template or custom project to edge and gateway layers with rollback by pointer.
Every capability through an organisation-scoped HTTP interface with an OpenAPI description.
Devices, inputs, streams, integrations, jobs, compute and keys in one place.
Tiers that set projects, services, compute, storage, models and rates, shown in your limits.
One sign-in for Living Worlds and Ipnops, with entitlements for each.
An address for every organisation and project on both sites.
Give your organisation's pages its own look within the product design.
Organise work as accounts, organisations and projects with roles and narrowing limits.
SQL, vector, time-series, graph, geospatial, search and key-value stores per project.
Set a project site to none, private or public on the same deployment.
Expiring, revocable keys per product and project, shown once and audited.
One data namespace, quota and audit trail shared by both products with explicit grants.
Roles, groups, service accounts and per-member database roles.
Send job and result events to your own systems.
Each kind of field survey has a typed intake slot. It checks units, ranges, time and height datum before a record set is accepted, and refuses anything that is not a real measurement.
Until a survey record arrives, the world runs on a labelled public stand-in. A record taken at the site replaces it, and the stand-in stays in the history.
Every capability ships with a recorded check you can read.
The capability lists on this site and in the documents are generated from the same source and checked on every change.
Get your API key.
Tell us what you want to run. We answer every request ourselves.