Compatible request formats
Use the OpenAI Chat Completions shape or the Anthropic Messages shape with the matching endpoint and credentials.
Unified AI inference gateway
One API.
Every model.
Connect your applications to multiple AI providers through a single, consistent API. Route requests across models, keep usage in one place, and switch providers without rewriting your integration.
OpenAI-compatibleAnthropic Messages-compatibleUsage ledger & prepaid credits
gateway-request.ts API preview
One endpoint, compatible request
POST https://your-gateway.example/v1/chat/completions
{
"model": "google/gemini-2.5-flash",
"messages": [
{
"role": "user",
"content": "Summarize this data..."
}
]
}
Configured provider examples
GPT-4.1 mini
openai/gpt-4.1-mini · OpenAI
Gemini 2.5 Flash
google/gemini-2.5-flash · Google
Claude Sonnet
anthropic/claude-sonnet · Anthropic
Example model IDs come from the starter configuration. Actual availability depends on provider credentials configured by the operator.
1 endpoint
for supported providers
2 API formats
OpenAI + Anthropic Messages
Per-request
usage and credit ledger
Vercel-ready
or self-host with Node.js
Model directory
Find the right model
for your workload.
Explore the model identifiers configured for this gateway. A model becomes callable after its provider credentials and pricing have been verified by the operator.
Loading model catalog…
Loading model catalog…
Model rates shown here are read from the starter provider configuration and are not a live quote from the upstream provider. Verify prices and context limits before production use.
Built for integrations
Less provider glue.
More time shipping.
A straightforward gateway layer for applications that need one integration surface across more than one model provider.
Central provider routing
Map public model IDs to upstream models in configuration, with ordered fallback among providers that share a public model name.
Usage and credit ledger
Track request usage and credit transactions per API key. Operators can create, fund, and revoke keys from the protected console.
Streaming responses
Forward server-sent event streams and convert basic text-message streams between supported API formats.
Keys stay server-side
Upstream provider secrets are configured as environment variables and are never included in public model metadata.
Deploy your way
Deploy the web console and API routes on Vercel, or run the Node.js server with PostgreSQL in your own environment.
Quickstart
Keep your client.
Change the base URL.
Create an account, generate a gateway API key from your dashboard, and send compatible requests through your gateway domain. API keys require credits before requests can run.
Create a user account to manage API keys and view usage. New keys start with zero credits; balance funding is currently handled by the operator.
Request example
Requires a gateway API key with available credits
Developer resources
Start with the endpoint
you already know.
OpenAI-compatible
Chat completions, bearer authentication, and server-sent event streaming.
View example ↗Anthropic-compatible
Messages API using the x-api-key header and anthropic-version.
Operations
Configure provider secrets, PostgreSQL, and the admin API key before serving traffic.
Open user dashboard ↗One integration. Clear usage records.
Explore the configured model catalog and integrate using either supported API format. The operator controls provider credentials and the initial API credit balance.