Product
The gateway platform
An OpenAI-compatible multi-model gateway. Requests are decided, executed, billed and recorded in this layer; this page states what each of the four parts is responsible for.
The endpoint
The protocol is OpenAI-compatible: change base_url and existing clients keep working. The service model name enthalpy-1 means the gateway chooses; a real pool model id skips routing and pins that model.
Decision records
Each call's candidate table, every attempt and every refusal reason go to route_decisions; the enthalpy_route object in the response carries the same information. Any decision can be retrieved in the console by request_id.
The ledger
Billing is itemised on actual upstream usage, to 1e-8 of a dollar; every entry expands to token counts, and reconciliation differences are listed separately.
The console
Key issuance and revocation, quotas and rates, usage rollups, decision and audit records, provided with the account.
4ROUTER / SYSTEMS RESEARCH
Connect configuration choices to system behavior.
Understand how operators, KV state and scheduling shape an inference path.
Read the systems notes in Resources ↗From operators to serving: cuBLAS & cuDNN
Identify the limiting resource before selecting the implementation.
↗STATE / CAPACITYKV cache: from tensor shapes to capacity budgets
Context length is a state budget that grows with concurrency.
↗SCHEDULING / SLOFrom prefill to decode: designing the scheduling experiment
When throughput rises, know which requests are waiting.
↗Let’s talk infrastructure.
Connect to the platform, or discuss models, performance and deployment.