First request

After Quick start, you should have a model route named fast and a gateway API key. These examples use the default address; adjust it if you changed the port.

export GPROXY_KEY='your-gproxy-api-key'

The gateway accepts Authorization: Bearer <key>, x-api-key, and x-goog-api-key. Use a gateway key, not an upstream credential.

OpenAI Chat Completions

curl -sS http://127.0.0.1:8787/v1/chat/completions \
  -H "Authorization: Bearer $GPROXY_KEY" \
  -H 'content-type: application/json' \
  -d '{"model":"fast","messages":[{"role":"user","content":"Say hello."}]}'

OpenAI Responses

curl -sS http://127.0.0.1:8787/v1/responses \
  -H "Authorization: Bearer $GPROXY_KEY" \
  -H 'content-type: application/json' \
  -d '{"model":"fast","input":"Say hello."}'

Claude Messages

curl -sS http://127.0.0.1:8787/v1/messages \
  -H "x-api-key: $GPROXY_KEY" \
  -H 'anthropic-version: 2026-10-21' \
  -H 'content-type: application/json' \
  -d '{"model":"fast","max_tokens":256,
       "messages":[{"role":"user","content":"Say hello."}]}'

Gemini GenerateContent

curl -sS "http://127.0.0.1:8787/v1beta/models/fast:generateContent" \
  -H "x-goog-api-key: $GPROXY_KEY" \
  -H 'content-type: application/json' \
  -d '{"contents":[{"parts":[{"text":"Say hello."}]}]}'

Streaming

curl -sSN http://127.0.0.1:8787/v1/chat/completions \
  -H "Authorization: Bearer $GPROXY_KEY" \
  -H 'content-type: application/json' \
  -d '{"model":"fast","stream":true,
       "messages":[{"role":"user","content":"Count to three."}]}'

Gemini streaming

curl -sSN "http://127.0.0.1:8787/v1beta/models/fast:streamGenerateContent?alt=sse" \
  -H "x-goog-api-key: $GPROXY_KEY" \
  -H 'content-type: application/json' \
  -d '{"contents":[{"parts":[{"text":"Count to three."}]}]}'

Responses use the protocol requested by the client. Success depends on route members, upstream capabilities, and conversion support. Changing the endpoint does not add capabilities to the upstream.

Model lists

Use the console’s model catalog to find configured gateway names. To query the caller-visible gateway catalog through the API:

curl -sS http://127.0.0.1:8787/portal/api/models \
  -H "Authorization: Bearer $GPROXY_KEY"

permitted indicates whether the caller can use an entry. GET /v1/models forwards the selected upstream’s model list, which is different from the gateway catalog. Application does not expose /portal/api over HTTP; use its in-app catalog.

Select a provider or namespace

URL	Model in request	Target
`/v1/chat/completions`	`fast`	The `fast` route
`/openai-main/v1/chat/completions`	Upstream model ID	The `openai-main` provider
`/acme/v1/chat/completions`	`fast`	The `acme/fast` route

See Models and routes for naming rules.

Common errors

Status	Meaning
`400`	Invalid request or configuration value
`401`	Missing, invalid, disabled, or expired key
`403`	Insufficient permissions or OAuth scope
`404`	Unknown model, route, or resource
`429`	A rate or quota limit applies
`5xx`	Gateway or upstream failure; inspect the logs

After a request, check usage and request records in the console. Body retention depends on logging settings; see Usage, logs, and audit.