Cost visibility your CFO will trust
The dashboard shows every request with its raw provider cost, your markup, and the model that served it. Export to XLSX for the monthly report. No guesswork about why the bill moved.
See the pricing calculator →OpenRouter pioneered the multi-model gateway. ScopeVeil is built for teams that want flat-rate transparent markup, a real cost dashboard, and the option to self-host. Here is the unvarnished comparison.
Sign up in 60 seconds. Top up via Stripe when you are ready.
baseURL and apiKey. Your OpenAI SDK does not change. Updated for what each platform actually shipped as of May 2026.
| Feature | ScopeVeil | OpenRouter |
|---|---|---|
| OpenAI SDK compatible | Yes | Yes |
| Pricing model Both are usage-based. ScopeVeil applies the markup per request; OpenRouter charges on the credit purchase. | Upstream cost + 5 to 15% markup | Upstream cost + platform fee on top-ups |
| Subscription required | No | No |
| Credit expiration | Never | Never |
| Native cost analytics | Built-in dashboard with per-request drill-down, XLSX export, cost charts | Activity log |
| Observability for your own keys | Yes, free up to 500k events/month via SDK | Not available |
| Privacy posture OpenRouter offers a no-logging route. ScopeVeil enforces a privacy boundary at the SDK layer for the observability product. | Operational metadata only; prompt content excluded by schema at ingest | Logging policy depends on the model provider |
| Self-hosted option | Yes, on Enterprise (full stack in your VPC) | Hosted only |
| Per-provider failover OpenRouter routes between providers automatically. ScopeVeil currently calls the provider you choose. | Roadmap | Yes |
| Provider catalog breadth If you need niche or obscure models, OpenRouter has more selection today. | OpenAI, Anthropic, Google, Mistral, Cohere, Groq, plus 9 more | 300+ models across many providers |
Highlights mark areas where ScopeVeil has a clearly different posture today. We will not pretend OpenRouter is bad. It is not. It is a different bet.
The dashboard shows every request with its raw provider cost, your markup, and the model that served it. Export to XLSX for the monthly report. No guesswork about why the bill moved.
See the pricing calculator →On the Enterprise tier the entire stack runs as a docker-compose bundle inside your own infrastructure. No request body ever crosses the boundary of your VPC. Useful when compliance or paranoia (or both) is the constraint.
Read the security posture →Already have provider keys you want to keep? Drop in the SDK and get the same dashboard for free up to 500k events per month. The gateway is optional. The observability is not a separate product with a separate login.
Quickstart in 60 seconds →If you are already on OpenRouter through the OpenAI SDK, here is the diff.
import OpenAI from 'openai';
const ai = new OpenAI({
- baseURL: 'https://openrouter.ai/api/v1',
- apiKey: process.env.OPENROUTER_API_KEY,
+ baseURL: 'https://gateway.scopeveil.com/v1',
+ apiKey: process.env.SCOPEVEIL_KEY,
});
// Same SDK. Same model strings. Same response shape.
const r = await ai.chat.completions.create({
model: 'openai/gpt-4o-mini',
messages: [{ role: 'user', content: 'hello' }],
});That is the whole migration. If you used OpenRouter through a wrapper, the wrapper does not need to change either.
Functionally yes for the common case (chat completions, streaming, tool calls, embeddings). Both expose an OpenAI-compatible HTTP API. If you rely on OpenRouter's automatic provider failover or a model not in our catalog yet, you may need to wait for parity.
Each request reserves an estimate, hits the upstream provider, then we charge the actual upstream cost plus your tier markup (5 to 15 percent). The dashboard splits the two so the markup is never hidden. Free tier is the highest markup; Enterprise the lowest.
Of course. The two systems do not conflict. Most teams point a single service or a single environment at ScopeVeil first and compare receipts before switching the rest.
The gateway adds a small fixed overhead on top of the upstream call. For streaming requests we forward bytes as they arrive, so time-to-first-token tracks the upstream provider closely.
For the hosted gateway, request and response bodies are read in-memory for token counting and discarded. We persist metadata only (model, tokens, latency, cost, your tags). For the self-hosted bundle, nothing leaves your VPC.
Sign up, generate a key, point one non-critical service at ScopeVeil for a week, compare the dashboard against your OpenRouter receipts. If the numbers match expectations, roll the rest.
Top up via Stripe and send your first request in under a minute.