Managed platform

Early access

Managed RLMesh runs your models against simulated environments on a hosted cluster, batching inference across rollouts. It works today, but it is still early. There are two ways in.

Already invited? Sign in, then follow your first evaluation.

Hosted beta

Run evaluations now

For researchers and teams who want to try it on their own models. Tell us what you're working on; we review requests individually and invite the people we approve to an organization of their own.

Prebuilt models and environments from the catalog
Your own model and environment images
Dashboard, Python SDK and CLI

Request access

Enterprise

Talk to us first

If you need agreed terms, a data-handling arrangement, or evaluations on infrastructure you operate, start with a conversation and we will work out the next steps with you.

Evaluations on a Kubernetes cluster you operate
Terms and data handling agreed with you
Capacity planned around your runs

What to expect from the beta

Requests are reviewed: Requesting access puts you on a waitlist. We review each request and email an invitation if we approve it. That can take a while, and not every request is approved.
It is early: The dashboard, API and CLI can change between releases, sometimes without notice. There is no uptime commitment or SLA.
Capacity is shared: Beta evaluations run on the RLMesh-hosted cluster alongside other teams, so large runs can queue.
Who can see your work: Your evaluations, images and results are visible to your organization's members and to RLMesh operators, who can read them to run and support the platform. Don't upload weights or data you are not comfortable sharing on those terms.
Access can change: We may limit, pause or end beta access while the platform takes shape.

Questions about any of this? Email research@competesai.com.