LLM gateway¶
The gateway is a single OpenAI-compatible endpoint that sits between your applications and every model behind them — self-hosted models on GRN.CLOUD GPUs, and external providers.
Without a gateway, every application carries its own provider keys, its own retry logic and its own idea of what a request costs. With one, those move to a single place your team can inspect, budget and audit.
What it handles¶
| Concern | What the gateway does |
|---|---|
| Keys | Issues virtual keys per team, app or user; provider keys stay server-side |
| Budgets | Per-key spend limits and rate limits, enforced before the request leaves |
| Routing | Sends a request to the right model by name, tag or policy |
| Failover | Falls back to a secondary provider or tier when the primary is unhealthy |
| Observability | Per-request logging of model, tokens, latency and cost |
| Compatibility | OpenAI-shaped API, so existing SDKs work unchanged |
The client's view¶
Because the API is OpenAI-compatible, most applications need a base URL change and nothing else:
from openai import OpenAI
client = OpenAI(
base_url="https://<your-gateway-host>/v1",
api_key="<your-virtual-key>", # issued by the gateway, not the provider
)
response = client.chat.completions.create(
model="<model-name>",
messages=[{"role": "user", "content": "Hello"}],
)
Where it sits¶
flowchart LR
APP1["App"] --> GW
APP2["Agent"] --> GW
APP3["Notebook"] --> GW
GW["LLM gateway"] --> SELF["Self-hosted<br/>vLLM on GRN GPUs"]
GW --> P1["External provider A"]
GW --> P2["External provider B"]
GW -.->|budgets · audit| ADMIN["Platform team"]
What this section will cover¶
- Gateway configuration and routing policy
- Virtual keys, budgets and rate limits
- Provider registration and failover tiers
- Request logging and cost attribution
Related¶
- Deploy an LLM gateway
- Add a user to the gateway
- Add a model provider
- AI inference — the models behind the gateway
Scaffold status
This is a placeholder. Configuration reference, routing policy syntax and the admin API are written here once confirmed against the platform.