Know and control AI spend before it surprises you.
Route OpenAI-compatible calls through MeterLayer, enforce budgets, and see token usage, cost, latency, errors, models, projects, customers, and prompts in one operational dashboard.
Log in to MeterLayer
Open your dashboard to review AI spend, logs, budgets, and gateway activity.
Why teams add a gateway
AI providers show a bill. MeterLayer shows what caused it.
The useful unit is not just provider spend. It is spend by product surface, customer, prompt, and operational state.
Know what every AI feature costs
Track spend by project, customer, feature, model, and prompt reference instead of waiting for a provider invoice.
Stop runaway usage automatically
Put monthly budgets, request limits, model allowlists, and provider controls in front of production traffic.
Keep provider keys out of apps
Store OpenAI credentials encrypted in MeterLayer and let teams call your project API key instead.
First traffic
Get the first production request through in minutes.
Use MeterLayer as a thin, observable control layer. Keep your app code close to the OpenAI SDK shape while gaining budgets and request logs.
Create a MeterLayer project and copy its project API key.
Store your OpenAI key once in Settings -> Providers.
Set your app base URL to https://www.meterlayer.io/v1 or call /api/proxy/chat directly.
Watch cost, tokens, latency, errors, budgets, models, and metadata arrive in the dashboard.
Use cases
Built for the moment AI usage becomes operational.
Start with one side project. Keep the same structure when customers, features, prompts, and costs multiply.
SaaS teams shipping AI features
Measure AI cost per tenant, customer, feature, and project before pricing gets painful.
Prompt-heavy products
Connect promptRef traffic to Prompt Registry and compare prompt cost, latency, and optimization signals.
Ops and finance handoff
Give teams request logs, budget alerts, audit events, CSV exports, and webhook delivery history.
Integration
Change the base URL. Keep your AI feature moving.
OpenAI-compatible clients can use https://www.meterlayer.io/v1. Gateway-native requests can use /api/proxy/chat with provider, model, metadata, and optional promptRef.
import OpenAI from "openai";
const client = new OpenAI({
apiKey: process.env.METERLAYER_PROJECT_KEY,
baseURL: "https://www.meterlayer.io/v1"
});
await client.chat.completions.create({
model: "gpt-4o-mini",
messages: [{ role: "user", content: "Write a welcome email." }],
max_tokens: 120,
metadata: {
feature: "email-generator",
customerId: "cus_123"
}
});Pricing
Simple plans while we validate the first customers.
Provider API costs are billed by your AI provider separately. MeterLayer pricing covers gateway, logs, budgets, analytics, and operational controls.
Starter
For side projects and early AI features.
$190/year when annual billing is enabled.
- 100k proxy requests / month
- 30-day log retention
- 3 projects
- Budget alerts
- OpenAI-compatible endpoint
Growth
For SaaS teams tracking customers, projects and features.
$790/year when annual billing is enabled.
- 1M proxy requests / month
- 180-day log retention
- 20 projects
- Metadata analytics
- Webhook alerts
Scale
For production teams that need stronger controls and auditability.
$2490/year when annual billing is enabled.
- 5M proxy requests / month
- 365-day log retention
- Unlimited projects
- Audit trail
- Ops readiness checks
FAQ
The practical questions before first launch.
Is MeterLayer replacing OpenAI?
No. MeterLayer sits in front of OpenAI as a gateway. Your app calls MeterLayer, MeterLayer forwards the request, captures usage and cost, and returns the provider response.
Do you store prompts?
Prompt content is not logged by default. MeterLayer stores usage, cost, latency, status, metadata, and promptRef context. Full prompt logging remains an explicit code-level option.
Where does Prompt Registry fit?
Prompt Registry is a separate product at prompts.meterlayer.io. MeterLayer can render promptRef requests through it and then show prompt-level analytics in the dashboard.
Can I start without Stripe?
Yes. The product can be used manually today. Stripe live mode is the remaining step for fully self-serve paid checkout.
Route one AI feature through MeterLayer this week.
Start with a low-risk feature, tag requests with metadata, and use the dashboard to understand real cost before scaling traffic.