Sample data. These figures are not a customer bill.Learn more
parezenparezen
Devbox Agents
Team · Sample · cloud
Home
  • Get startedSoon
  • Audits
  • Overview
Analyse
  • Spend
  • Cost per taskLock
  • Agents
  • Model comparisonSoon
  • Budgets
  • AlertsSoon
  • Reports
  • PR cost checksSoon
  • AdvisorsSoon
  • Spend analysis
Optimise
  • RecommendationsSoon
  • Routing policiesSoon
  • Shadow testsSoon
  • CachingSoon
  • TrimmingSoon
  • BatchSoon
  • Waste detectorSoon
Gateway
  • SetupSoon
  • ProvidersSoon
  • ModelsSoon
  • API keysSoon
  • DeploymentsSoon
  • PlaygroundSoon
  • LogsSoon
Savings & billing
  • Savings history
  • Savings ledger
  • Bill match
  • Billing
  • Invoices
Settings
  • Data & privacySoon
  • OrganisationSoon
  • ProjectsSoon
  • TeamSoon
  • RolesSoon
  • SSOSoon
  • Audit logSoon
Plan: Team · Sample ↗
DocsHelp

Sample data. Not a customer bill.

Devbox Agents/Routing policies/Policy editor

Policy editor

Planned

Right-size, cascade, cache, batch, automation, YAML and history.

Included with Scale

Policy editor is plan-gated in this sample. The preview stays visible.

Compare plans

Code review · review_posted · version 3

Thresholds

The customer's bar. A lever that changes quality cannot cross it without a passed shadow test.

appCode review agentgatewayGatewayconditionshadow test passedmodelopenai/gpt-4.1-minifallbackopenai/gpt-4.1-minifallbackopenai/gpt-4.1provideropenai
  1. app
    Code review agent
  2. gateway
    Gateway
  3. condition
    shadow test passed
  4. model
    openai/gpt-4.1-mini
  5. fallback
    openai/gpt-4.1-mini
  6. fallback
    openai/gpt-4.1
  7. provider
    openai
Shadow testsAll tests for this routeSavingsLogs

What this policy will do

  • Code review agent follows a latency-first policy.
  • A served response has to score at least 0.84 and finish inside 450 ms at p95.
  • Anything priced above $0.0150 a request is outside this policy.
  • Simple requests can move from openai/gpt-4.1 to openai/gpt-4.1-mini only after a shadow test passes.
  • Cascade is off, so there is no cheap-first attempt.
  • If the chosen model fails, fallbacks run in order: openai/gpt-4.1-mini, openai/gpt-4.1.
  • Output is capped at 600 tokens, with reasoning effort low.
  • The route is synchronous, so it stays off batch prices.
  • Exact-match response cache is off for this route.
  • Semantic cache is Planned and is not applied.
  • Parezen only suggests. A person still chooses Enable or Ignore.

Simulated impact

12 sample requests on this route.

Baseline
$0.03
Actual
$0.01
Estimated spend
$0.01est.▲ 0.0%
Measured p95
765 ms
Escalations
0
Fallbacks
2
  • This sample has no cascade escalations. When one happens, its cost is included, not hidden.
  • Exact-match cache is off, so responses are not reused.
  • Semantic cache is Planned and is excluded.