Docs

Short reference for your local OpenAI-compatible gateway.

Base URL

https://api.f1shy312.com

Use this for direct gateway calls.

OpenAI-style URL

https://api.f1shy312.com/v1

Use this as SDK base URL.

Auth

Personal mode

Bearer keys are optional locally.

Endpoints

Method	Path	Explanation
GET	/health	Gateway status, base URL and prompt logging state.
GET	/v1/models	OpenAI-compatible model list with your configured aliases.
GET	/v1/models/:model	Details for a single model alias.
POST	/v1/chat/completions	OpenAI-compatible chat endpoint. Supports streaming with SSE.
POST	/v1/responses	Responses API for Codex-style clients. Supports streaming and stored responses.
GET	/v1/responses/:response_id	Retrieves a stored response by ID.
DELETE	/v1/responses/:response_id	Deletes a stored response.
POST	/v1/responses/:response_id/cancel	Cancels an in-progress response.
GET	/v1/responses/:response_id/input_items	Lists the input items of a stored response.
GET	/admin/stats	Dashboard metrics for requests, latency, errors and cost.
GET	/admin/providers	Configured providers, type, base URL and enabled state.
POST	/admin/providers	Creates a new provider.
POST	/admin/providers/test	Checks whether a provider endpoint is reachable.
PATCH	/admin/providers/:id	Updates a provider's name, type, base URL or enabled state.
DELETE	/admin/providers/:id	Deletes a provider.
GET	/admin/models	Model aliases, provider routes, fallbacks and basic route metrics.
POST	/admin/models	Creates a new model alias route.
PATCH	/admin/models/:id	Updates a model alias route.
DELETE	/admin/models/:id	Deletes a model alias route.
POST	/admin/models/test	Tests a model alias by sending a minimal request through its route.
GET	/admin/api-keys	Lists locally generated gateway keys without exposing secret values.
POST	/admin/api-keys	Creates a gateway key and reveals it once.
PATCH	/admin/api-keys/:id	Enables or disables a gateway key.
DELETE	/admin/api-keys/:id	Deletes a gateway key.
GET	/admin/logs	Metadata-only request logs. Prompts and responses are not stored.
GET	/admin/settings	Runtime settings shown in the dashboard.
GET	/admin/quota	Z.ai quota policy summary plus local quota/rate-limit events.
GET	/admin/quota-settings	Per-provider and per-model quota configuration.
PATCH	/admin/quota-settings	Updates quota limits, windows and enabled state.

Chat completion

curl https://api.f1shy312.com/v1/chat/completions \
  -H "Content-Type: application/json" \
  -d '{
    "model": "glm5.1",
    "messages": [
      { "role": "user", "content": "Write a small TypeScript function." }
    ],
    "temperature": 0.2
  }'

Streaming

curl -N https://api.f1shy312.com/v1/chat/completions \
  -H "Content-Type: application/json" \
  -d '{
    "model": "glm5-turbo",
    "messages": [
      { "role": "user", "content": "Stream a short coding answer." }
    ],
    "stream": true,
    "stream_options": { "include_usage": true }
  }'

Responses API

curl https://api.f1shy312.com/v1/responses \
  -H "Content-Type: application/json" \
  -d '{
    "model": "glm5-turbo",
    "instructions": "You are a coding assistant.",
    "input": "Inspect this error and suggest a fix.",
    "stream": false
  }'

For Codex-style clients. Stored responses can be retrieved by ID unless store is set to false.

Use it like a normal OpenAI-compatible API

baseURL: "https://api.f1shy312.com/v1"
model: "glm5.1"
apiKey: "any-local-value-or-your-generated-key"

Provider keys stay in the gateway. Client tools only need the local base URL and a model alias.