Deployment environment
Bring your on-premises and infrastructure requirements into the planning conversation, alongside the model and the expected demand.
Edge operates production inference in controlled environments.
Define the deployment and operating requirements alongside the model’s quality, performance, and cost targets. Edge runs the same operating loop as Stack, with explicit boundaries on what can run automatically, and keeps production request and response payloads in your environment.
What Edge does for your organization
Edge is Sparset’s inference operations edition for organizations with stricter governance, control, and on-premises requirements. It brings planning, optimization, deployment, operation, and ongoing review into one operating loop, with recovery limits your team defines and your approval on every optimization and configuration change. The actual architecture, connectivity, supported controls, and commercial commitments are established for your deployment during evaluation, not assumed.
The requirements your deployment has to meet
The model is one part of the decision. Edge is for organizations that also need to establish where inference runs and how production changes are controlled.
Data handling
Production request and response payloads stay in your environment. Review the deployment’s operational data and access requirements with Sparset.
Operating authority
Authorize, at setup, which of worker restarts, traffic shifts, failover, and rollback can run, and within what limits. Keep approval of optimization and configuration changes with your team.
From first deployment to everyday operation
Sparset handles the operating loop. Your team sets the requirements and approves production changes.
Plan
Assess your model and workload to choose hardware and a serving configuration. Your team defines requirements and targets.
Optimize
Tune and test the configuration against quality, performance, reliability, and cost criteria. Your team agrees on what the deployment must achieve.
Deploy
Put the tested configuration into production after approval. Your team approves the production change.
Operate
Monitor continuously and run authorized recovery within your limits. Your team sets recovery permissions and boundaries.
Review
Review production at least every 24 hours and test potential improvements. Your team accepts or rejects proposed changes.
Control continues after deployment
Four things that happen in production, and what happens with Edge.
A worker stops responding
Monitoring can trigger a restart if the action is authorized at setup and within the agreed limits. The action leaves the record of what ran and why.
An optimization is found
The proposal is tested against agreed quality, performance, reliability, and cost criteria. Your team reviews the benefits and tradeoffs before approving a change.
A new model is proposed
Planning, optimization, and validation run against the proposed model. Your team sees the tested comparison before approving promotion.
A workload needs its own limits
Models run as separate jobs, each with its own targets, limits, and approval rules.
Edge or Stack
Choose the edition around your requirements. Both run the same five-stage loop, with the same recovery limits and approvals.
Edge
For organizations with stricter governance, control, and on-premises requirements: start with the deployment environment and operating requirements alongside the workload.
Stack
For engineering and platform teams that want help planning, deploying, and operating production inference: start with the model, the workload, and the GPU capacity you have or need.
Questions to resolve before deployment
The true answers, including where the answer is “bring it to the evaluation.”
When should we choose Edge instead of Stack?
Evaluate Edge when stricter governance, operating control, or on-premises requirements are central to the deployment. Stack focuses on the inference operating needs of engineering and platform teams. We discuss the requirements with you to determine fit.
Is Edge intended for on-premises deployments?
Yes. On-premises requirements are part of Edge’s focus. Bring your infrastructure, network restrictions, and workload to the evaluation so support for the proposed deployment can be confirmed.
Do you access or store our production data?
Production request and response payloads stay in your environment. We walk through the operational data and access required for the proposed deployment during evaluation.
Can Edge operate without internet connectivity?
If disconnected operation is required, raise it during evaluation so support can be confirmed before the deployment is agreed.
What can run without a fresh approval?
Only recovery actions authorized at setup and within your limits: worker restarts, traffic shifts, failover, and rollback. Optimization and configuration changes require approval.
Does Edge replace our platform or security team?
No. Edge handles the inference operating loop. Your organization still defines its requirements, sets recovery limits, and approves optimization and configuration changes.
Can Edge meet our required controls and service commitments?
Bring the specific requirements to the evaluation. We review what the proposed deployment can support and what commitments can be agreed before you proceed.
How is Edge priced?
Pricing depends on the edition, deployment scale, and whether you bring GPU capacity. Book a call and we will walk through it against your workload.