Skip to main content
Once an AI feature reaches production, you need to understand what it is doing and change its operating boundaries safely. Observe & Control brings those tools together in the Opper platform.

Observe what happened

Logs

See every provider call with its model, provider, status, tokens, cost, and latency, on every plan.

Traces

Inspect the model call, nested tool calls, cost, latency, and rule activity behind a request.

Files

Manage files used as model inputs and generation outputs.

Control what is allowed

Rules is the organization-level workspace for the policies Opper enforces on model calls.

Opper retention

Turn Opper tracing on or off and choose how long to keep traces.

Spend limits

Set a monthly organization budget and optional project caps.

Model access

Restrict the providers, regions, countries, and models your apps can use.

Checks

Score quality or flag, redact, and block content.

Routing

Choose which provider a model pool tries first.
Default-model and response-caching rules are set through the Management API:

Default model

Set a default model for an organization or project.

How rules apply

Organization rules set the default for every project. Project overrides can tighten that default, but cannot loosen it.
  • Model-access rules intersect, so a model must be available at both levels.
  • The shortest applicable retention period wins.
  • Project spend caps must fit inside the organization budget.
  • Checks can run across the organization or only on selected projects.
When trace storage is enabled, rule activity appears on the relevant trace. In the playground, turn on Project controls to test a call with its rules applied.
Last modified on October 2, 2026