Usage and model guardrails

Keep AI access useful without letting spend run loose.

OpenCentric meters local trial activity, routes anonymous users to capped preview access, and reserves advanced models for paid plans with protected margins.

Local browser meter

Your current preview usage.

Create account

Plan Free browser preview

Usage loads from this browser.

Messages 0 / 3

Tokens 0 / 10,000

Access value $0.000 / $0.350

Local storage preview

This page shows what this browser has used before sign-up. Production model spend should be enforced by the OpenCentric backend with server-side quotas, abuse checks, billing status, model routing, and hard caps.

AnonymousCapped preview only.

Use low-cost model access, short outputs, no advanced tools, and require sign-up to save or continue.

Free accountMonthly route allowance.

Let users save sessions, continue ideas, and see usage while advanced models remain locked.

Paid usersAdvanced models with budget rails.

Paid tiers unlock deeper architecture, compliance, and cost analysis while keeping a 60% target gross margin.

Paid tier economics

Price model access so OpenCentric keeps 60% margin.

View pricing

Guardrail stack

Human-only access needs layers, not a single switch.

Human checkUse Turnstile or equivalent before sensitive model calls.Required

AI gatewayNever expose model keys in the browser. Route every request through OpenCentric.Server-side

Quota ledgerTrack messages, tokens, cost, model, account status, and remaining access.Persistent

Model routerAnonymous users get mini-model previews. Paid users unlock advanced models by plan.Controlled

Hard capsStop calls when user, workspace, plan, or abuse limits are reached.Protected

Recent local activity

Think Session usage saved in this browser.

Estimated usage is for the local preview experience. Final production billing should use server-side token and cost records.