Reliability Studio / Pricing
Pay for operational depth as your reliability program grows. Run on infrastructure you control, or fund managed execution separately—with a visible estimate and hard limit before launch.
LAUNCH STATUS / Studio and customer-hosted execution are available in founding access. Iso Compute is a capacity-gated private beta and is never assumed available.
Plans
Every plan can inspect the complete evidence behind a result. Pro adds guided building, code inspection, and truthful simulation rendering. Team adds the full collaborative workspace, editable manifests, advanced perturbation views, and failure replay.
Community
$0/ month
For learning the complete reliability loop on one active project.
Founding Pro
$79/ month
Limited to the first 100 verified Pro workspaces.
Founding Team
$249/ month · first year
Limited to the first 10 paid Team workspaces.
See exactly how each paid plan works in Studio: compare Pro and Team Build & Run workflows · review rendering requirements · understand workspace access →
FOUNDING PRO / TERMS
The first 100 verified Pro workspaces receive two promotional managed worker-hours and the option to pay $79/month for their first 12 paid months. The grant expires with the 30-day trial. After the first 100, the software trial remains free, managed CPU is prepaid, and standard Pro is $149/month. Nothing charges automatically.
FOUNDING TEAM / TERMS
The first 10 paid Team workspaces receive $249/month for their first 12 paid months; standard Team is $599/month. Three users are included. Users 4–10 are $99 per user per month and each adds 60 managed worker-hours to the shared monthly pool.
Compare
| Capability | Community | Pro | Team |
|---|---|---|---|
| Complete evidence | Included | Included | Included |
| Users | 1 | 1 · fixed | 3 included · users 4–10 +$99 each |
| Workspace | Personal · 1 active project | Personal · multiple active projects | Shared · multiple active projects |
| Build & Run | — | Guided configuration | Full workspace |
| Code access | SDK + CLI | Generated Python + JSON | Generated code + editable manifest |
| Environment rendering | — | Declared + recorded | Declared + recorded |
| Scenario perturbation views | — | Standard | Advanced |
| Failure replay | — | — | Included |
| Active projects | 1 | Multiple | Multiple |
| Scheduled suites | — | Included | Included |
| Reliability gates | — | Included | Included |
| Integrations | Core | GitHub + CI/CD | Advanced |
| Heavy-artifact retention | 7 days | Target 90 days | Configurable |
| Roles + approvals | — | — | Included |
| Managed worker-hours | BYOC | 60/month | 240 + 60 per added user |
| Managed execution | 1 CPU job | 1 CPU job | 1 CPU job · higher limits |
Compute / private beta
Iso Compute provisions one dedicated ephemeral CPU worker, runs your authorized container, meters it, and shuts it down. GPU and distributed execution stay in infrastructure you control.
PATH 01 / BRING YOUR OWN
Run locally or in infrastructure you control and send only the evidence and telemetry you authorize.
ISO COMPUTE USAGE / $0PATH 02 / ISO COMPUTE
Use included or purchased worker-hours, review the estimated runtime, then run one isolated 4-vCPU worker at a time.
PER-SECOND METERING / ONE-MINUTE MINIMUMISO COMPUTE / WORKLOAD SIZING
One worker-hour means one isolated 4-vCPU / 16 GB worker for one hour. GPU training and distributed workloads are bring-your-own-compute only.
| Workload | Comparative size | Execution | Guidance |
|---|---|---|---|
| Local smoke check | 1–4 short runs | Laptop / CI | Local recommended |
| Managed simulation | 5–60 typical 2-minute runs | 4 vCPU · 16 GB | Pro |
| Managed batch | Up to ~120 typical 2-minute runs | 4 vCPU · 16 GB | Team |
| RL policy evaluation | Rollouts and inference checks | 4 vCPU · 16 GB | CPU suitable |
| Neural training / world models | Training or distributed jobs | Customer GPU | BYOC |
USAGE VISIBILITY / EXAMPLE
ON-DEMAND AVAILABILITY / EXAMPLE
Iso Compute uses best-effort on-demand capacity. Availability varies by region and demand. No worker-hours are consumed while queued or when placement fails, and Studio never substitutes a more expensive worker without approval. Reserved capacity is not included at launch.
Other paths
RESEARCH ACCESS
Approved academic, nonprofit, and independent researchers receive time-bound Pro access and one nonrenewing grant of 100 managed worker-hours.
Apply for Research AccessDESIGN PARTNERSHIP
Bring one meaningful system and work directly with us to establish a connected, repeatable reliability program.
Discuss a pilotClear terms
No. At the end of 30 days, you choose whether to begin a paid subscription or return to Community.
The first 100 verified Founding Pro workspaces receive two promotional worker-hours during the trial. Later trials use BYOC or purchased hours. Paid Pro includes 60 managed worker-hours each month and Team includes 240 plus 60 for each paid added user. One worker-hour is one isolated 4-vCPU / 16 GB worker for one hour.
Yes, through a customer-hosted runner on GPU capacity you control. Iso receives a short-lived, scoped runner credential; your cloud-provider keys stay in your secret manager and never pass through Iso.
CPU capacity is broadly available and predictable enough for a prepaid beta. Keeping GPU in customer infrastructure avoids hidden substitutions, quota surprises, and custody of customer cloud credentials.
No. Founding Pro is a single-user plan and does not support additional users. Founding Team starts with three included users and 240 shared managed worker-hours. Users 4 through 10 cost $99 per user per month, and each adds 60 worker-hours to the shared monthly pool.
Yes. Scope, uncertainty, limitations, disposition, and provenance remain visible. Pricing does not hide the truth of an existing result.
They are launch pricing intended to be validated during beta. Founding participants who opt in receive the stated first-year rate.