Pricing
Spend cents to protect every GPU dollar
One product, priced by the GPU capacity it covers. Standard managed plans use graduated rates, include the complete investigation workflow, and begin at $40 per covered GPU per month ($0.055 per covered GPU-hour).
Graduated fleet pricing01 02 03
GPUs 65–128
$20per covered GPU / month($0.027 per covered GPU-hour)Above 128 GPUs
Customfleet, site, or OEM agreementPricing reflects deployment and tenant coverage.| Covered fleet | Effective RidgeScope price | RidgeScope / $3.85 GPU-hour |
|---|---|---|
| 64 GPUs | $40 / GPU / month ($0.055 / GPU-hour) | 1.4% |
| 128 GPUs | $30 / GPU / month ($0.041 / GPU-hour) | 1.1% |
Recovered GPU time
What is 5% useful GPU time worth?
Compare recovered compute value with RidgeScope's per-GPU price.
Public on-demand rate from Nebius.
Share of total monthly GPU time recovered.
Illustrative recovered compute value$1,124per 8-GPU node / month$140.53 per GPU / month
8 GPUs × 730h × $3.85 × 5%
Everything in one product01 02 03 04
No feature maze between detection and action
RidgeScope is valuable when fleet context and job evidence stay connected. The managed plan includes the complete workflow instead of splitting it across monitoring, investigation, and user tiers.
Fleet Mission Control
Productive, occupied-idle, free, and unhealthy GPU capacity—measured across the fleet and priced at your own GPU rate.
Cited AI investigations
Job-level verdicts that connect scheduler state, GPU execution, distributed progress, logs, and the training outcome.
The full evidence layer
GPU, CUDA, NCCL, Slurm, host, storage, fabric, and job-output telemetry without an SDK or training-code change.
Capacity-based access
Pricing follows covered GPU capacity—not seats, dashboards, users, or individual investigations.
The commercial meter is covered GPU capacity—not seats or individual investigations.
Deployment and scaleStandard managed Private deployment Neocloud / OEM
The same evidence model, three commercial paths
For AI labs and ML platform teams
RidgeScope operates the data and investigation path. Coverage begins at $40 per covered GPU per month ($0.055 per covered GPU-hour) and steps down automatically with fleet size.
Discuss managed pricingFor on-premises and air-gapped environments
Keep telemetry, secrets, data stores, and investigation inference inside your security boundary. Coverage and deployment are quoted together.
Discuss private deploymentFor fleet-wide and embedded coverage
Use a negotiated platform and covered-tenant agreement instead of applying the standard linear rate across resale capacity.
Discuss fleet pricingStart with evidenceFree Fixed scope · credited
Audit first. Deploy when there is something worth proving.
GPU utilization audit
Review a fleet snapshot or difficult training run. We identify the clearest cost exposure and whether a deployment is likely to add useful evidence.
Request the free auditDeployment-based proof of value
A fixed-scope engagement on one cluster with success criteria and commercial terms agreed up front. The fee is credited when you continue.
Scope a proof of valuePricing FAQ
The meter, commitment, and fit
What is a covered GPU-hour?
One covered GPU-hour is one physical GPU included in RidgeScope coverage for one clock hour, whether it is computing, waiting, idle, or unhealthy. The published hourly figure is the equivalent rate of committed fleet coverage.
Do you bill only when a training job is active?
No. Billing only active or productive time would make RidgeScope cheaper when waste is high and punish it for helping the fleet improve. Contracts cover a baseline fleet; elastic capacity above that baseline can be handled as hourly overage.
Why is there a 64-GPU minimum?
RidgeScope includes cluster-wide telemetry, deployment, data infrastructure, and investigation support. The minimum keeps the managed plan viable while covering a fleet large enough for the recovered-capacity economics to be meaningful.
Does the price change for H100, H200, or B200 GPUs?
No. The RidgeScope coverage rate is hardware-neutral. More expensive GPUs naturally strengthen the ROI comparison without a separate hardware-class surcharge.
Are users or investigations metered separately?
No. The standard commercial meter is covered GPU capacity, not seats, dashboards, or individual investigations.
How are fleets above 128 GPUs priced?
Large AI labs, neoclouds, and embedded deployments use a negotiated fleet or site agreement based on covered capacity, deployment model, support, and tenant use.
Can we validate RidgeScope before a fleet agreement?
Yes. Start with a free utilization audit. A deployment-based proof of value has a fixed scope and can be credited toward the fleet agreement if you continue.
