Solo
$ 39
/month
For individual engineers and researchers getting real work done.
- Component
- Paper
- 1 personal workspace
- Community model library
- 20 compute-hours / month
- 5 GB personal storage
- Standard support
Pricing
Scale by usage. Start on a seat, add compute when a run needs it, and stop paying for it when the run ends.
Team
$ 99
/user · month
For teams shipping engineering work together on all four products.
- Component
- Paper
- Exploration
- Design
- Real-time shared workspaces & reviews
- Version history, roles & permissions
- 200 compute-hours / month pooled
- 100 GB shared team storage
- Priority support
Enterprise
Custom
For organizations with scale, security and compliance needs.
- Everything in Team
- SSO / SAML & audit logs
- Dedicated or on-prem cluster
- Custom compute & storage commitments
- Private model registry
- SLA & solution engineering
Fine print
All plans include browser-native modeling, the community model library and standard support. Prices shown in USD, excluding tax. Annual billing saves 20%.
Add-ons
Attach these to any plan. Compute is billed by usage as you spin up more of the cluster; storage is a simple monthly subscription for your shared team space.
Compute
Usage-based
from $ 0.12
/vCPU-hour
Scale past your plan's included hours by spinning up more of the cluster for large sweeps and optimization runs — pay only for what you use.
- Autoscaling cluster nodes
- GPU nodes available on demand
- Per-second metered billing
- Set spend caps per workspace
Storage
Subscription
$ 8
/100 GB · month
Expand shared team space for models, runs and results with an addressable, versioned store that grows with you.
- Shared, versioned team space
- Addressable, reproducible records
- Encrypted at rest
- No egress fees
Every plan
The parts that are not metered
Some things do not belong behind a tier. Whichever plan you are on, these work the same way.
Compute
Metered by the second, not by the seat.
An idle model costs nothing. A sweep costs what it used, and the meter stops when the last run does.
A metering panel over the run queue that produces it. The session total climbs in core-seconds while two runs hold six cores between them, the queued sweep takes the last two on its own, and when it returns the rate falls to zero and the total stops moving — no runs, meter stopped.
Support
Answered by people who know the solver.
Support is staffed by the engineers who build ODE, not by a tier of scripts in front of them.
A support thread whose reply is solver-level. An initialisation question that stalls at a residual of 1e-3 is picked up by a named engineer, and the answer names the state variable it parks on — pump.port_b.h_outflow — then the debug flag and the 1e-8 tolerance to run it again with.
Security
The same isolation on every plan.
Encryption, tenant isolation and audit trails are not upsells. Only SSO and retention limits move with the tier.
A tenant panel that holds still while its audit trail appends. Four workspaces sit inside one named boundary with a neighbouring tenant dimmed and unnamed beside them, AES-256-GCM at rest and TLS 1.3 in transit are stated as configuration, and two entries — a key rotation and a run read — arrive newest-first on an append-only log.
Trusted by engineering teams building the physical world
- German Aerospace Center (DLR)
- Technical University of Munich
- LTX
- Volkswagen
- Bühler
- Ekium
- GEMÜ
- Origen