Call leading models through one API. ModelRelay meters every request to the customer who made it.
Provider rates pass straight through at a 0% ModelRelay fee, and every new account starts with $1 in credit.
Models · API · Monetize
Speed, cost, intelligence.
Tune the balance between speed, cost, and intelligence, then inspect the same model you can call through the API.
Every model runs the same controlled task, with output and performance you can inspect.
See serving speed, provider pricing, and independently sourced intelligence together.
Move to a different model by changing one string, not by rebuilding around another provider.
One request shape.
Use ModelRelay's provider-neutral Responses API, or point an existing OpenAI or Anthropic integration at the compatible endpoints.
See how requests flowcurl https://api.modelrelay.ai/api/v1/responses \
-H "Authorization: Bearer $MODELRELAY_API_KEY" \
-H "Content-Type: application/json" \
-d '{"model":"claude-sonnet-5","input":"Hello"}'Change the model. Keep everything else.
Usage in. Revenue out.
Know who made each request, what it cost, and what their plan allows.
Use customer tokens to attribute requests and usage to the right end user.
Define plans, approved models, credits, and spend limits in one control plane.
Set customer pricing while ModelRelay keeps the underlying provider cost attributable.
ModelRelay combines model access, request controls, usage accounting, and customer billing in one hosted API.
Validate the project or customer token.
Send the request to the model's provider.
Record returned usage and available provider cost.
Apply usage to the correct customer and plan.
ModelRelay is invite-only while we onboard early teams. Pay by card or USDC.