LLM Gateway Admin

Overview

The LLM Gateway Admin API provides observability and policy controls for organization owners and admins. It covers usage aggregation, request logs, routing decisions, DLP rules, budgets, and inference hooks. All admin routes require a Clerk JWT and organization owner or admin role.

Base URL

https://api.allternit.com/api/v1/gateway

For local development:

http://127.0.0.1:8013/api/v1/gateway

Endpoints

Method	Path	Description
`GET`	`/api/v1/gateway/usage`	Usage aggregation by day, provider, model, and key
`GET`	`/api/v1/gateway/logs`	Request logs with filtering and cursor pagination
`GET`	`/api/v1/gateway/routing/decisions`	Routing decisions for policy-alias requests
`GET` / `PUT`	`/api/v1/gateway/routing/policies`	List or update routing policies
`GET` / `PUT`	`/api/v1/gateway/dlp/rules`	List or update DLP rules
`GET` / `PUT`	`/api/v1/gateway/budgets`	List or update budgets
`GET` / `PUT` / `POST` / `DELETE`	`/api/v1/gateway/inference-hooks`	Manage inference hooks
`POST` / `GET` / `PATCH` / `DELETE`	`/api/v1/gateway/keys`	Manage virtual keys

Virtual-key management is documented on the Authentication page.

Authentication

curl https://api.allternit.com/api/v1/gateway/usage \
  -H "Authorization: Bearer $ALLTERNIT_TOKEN"

Usage

Query parameters

Parameter	Type	Description
`from`	string	Start date in `YYYY-MM-DD` format.
`to`	string	End date in `YYYY-MM-DD` format.

Example request

curl "https://api.allternit.com/api/v1/gateway/usage?from=2026-08-01&to=2026-08-13" \
  -H "Authorization: Bearer $ALLTERNIT_TOKEN"

Example response

{
  "from": "2026-08-01",
  "to": "2026-08-13",
  "usage": [
    {
      "day": "2026-08-13",
      "provider_id": "provider-a",
      "model_id": "model-x",
      "virtual_key_id": "key_abc123",
      "key_prefix": "ak-prod",
      "requests": 1240,
      "prompt_tokens": 450000,
      "completion_tokens": 120000,
      "reasoning_tokens": 8000,
      "spend_microdollars": 1250000,
      "cost_mismatches": 2
    }
  ],
  "totals": {
    "requests": 1240,
    "prompt_tokens": 450000,
    "completion_tokens": 120000,
    "spend_microdollars": 1250000
  }
}

Logs

Query parameters

Parameter	Type	Description
`limit`	integer	Page size. `1`–`200`. Defaults to `50`.
`cursor`	string	Pagination cursor in the form `created_at	id`.
`status`	string	Filter by request status.

Example request

curl "https://api.allternit.com/api/v1/gateway/logs?limit=10&status=completed" \
  -H "Authorization: Bearer $ALLTERNIT_TOKEN"

Example response

{
  "logs": [
    {
      "id": "evt_abc123",
      "status": "completed",
      "error_type": null,
      "policy": "allternit-balanced",
      "provider_id": "provider-a",
      "model_id": "model-x",
      "fallback_from": null,
      "prompt_tokens": 120,
      "completion_tokens": 45,
      "cost_microdollars": 1250,
      "latency_ms": 890,
      "ttft_ms": 120,
      "gizzi_session_id": "sess_abc123",
      "key_prefix": "ak-prod",
      "created_at": 1723500000
    }
  ],
  "next_cursor": "1723500000|evt_abc123"
}

Routing decisions

Query parameters

Parameter	Type	Description
`limit`	integer	Page size.
`cursor`	string	Pagination cursor.

Example request

curl "https://api.allternit.com/api/v1/gateway/routing/decisions?limit=5" \
  -H "Authorization: Bearer $ALLTERNIT_TOKEN"

Example response

{
  "decisions": [
    {
      "id": "rd_abc123",
      "policy": "allternit-code",
      "winner": {
        "provider_id": "provider-a",
        "model_id": "model-x"
      },
      "fallbacks": [
        {"provider_id": "provider-b", "model_id": "model-y"},
        {"provider_id": "provider-c", "model_id": "model-z"}
      ],
      "scores": [
        {
          "provider_id": "provider-a",
          "model_id": "model-x",
          "quality": 0.92,
          "cost_efficiency": 0.78,
          "total": 0.89,
          "heuristic": "benchmark_weighted",
          "benchmarks": {"humaneval": 0.91, "mmlu": 0.88}
        }
      ],
      "rules_fired": ["allternit-code"],
      "served_provider_id": "provider-a",
      "served_model_id": "model-x",
      "status": "served",
      "usage_event_id": "evt_abc123"
    }
  ],
  "next_cursor": "1723500000|rd_abc123"
}

Routing policies

List policies

curl https://api.allternit.com/api/v1/gateway/routing/policies \
  -H "Authorization: Bearer $ALLTERNIT_TOKEN"

{
  "policies": [
    {
      "name": "allternit-balanced",
      "weights": {
        "humaneval": 0.20,
        "mmlu": 0.20,
        "gpqa": 0.15,
        "gsm8k": 0.15,
        "arena_elo": 0.15,
        "aider": 0.15
      },
      "source": "preset"
    }
  ]
}

Update a policy

curl -X PUT https://api.allternit.com/api/v1/gateway/routing/policies \
  -H "Authorization: Bearer $ALLTERNIT_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{
    "name": "custom-code-policy",
    "weights": {
      "humaneval": 0.50,
      "mmlu": 0.10,
      "aider": 0.40
    }
  }'

Budgets

List budgets

curl https://api.allternit.com/api/v1/gateway/budgets \
  -H "Authorization: Bearer $ALLTERNIT_TOKEN"

{
  "budgets": [
    {
      "id": "budget_abc123",
      "scope": "organization",
      "budget_cents": 100000,
      "hard": true,
      "current_month_spend_microdollars": 45230000
    }
  ],
  "tenant_id": "tenant_abc123",
  "current_month_spend_microdollars": 45230000
}

Update a budget

curl -X PUT https://api.allternit.com/api/v1/gateway/budgets \
  -H "Authorization: Bearer $ALLTERNIT_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{
    "budget_cents": 150000,
    "hard": true
  }'

A hard budget blocks requests when exceeded. Soft budgets are used for reporting and alerts.

Inference hooks

Inference hooks are documented in detail on the Security: Inference Hooks page. Admin endpoints:

Method	Path	Description
`GET`	`/api/v1/gateway/inference-hooks`	Get hook configuration
`PUT` / `POST`	`/api/v1/gateway/inference-hooks`	Create or update hooks
`DELETE`	`/api/v1/gateway/inference-hooks`	Remove hooks

DLP rules

DLP rules are documented on the Security: DLP page. Admin endpoints:

Method	Path	Description
`GET`	`/api/v1/gateway/dlp/rules`	List rules and default action
`PUT`	`/api/v1/gateway/dlp/rules`	Create or update a rule

Error codes

HTTP status	Code	Meaning
400	`allternit.invalid_request`	Malformed query parameter or request body.
403	`allternit.insufficient_role`	Caller is not an owner or admin.
404	`allternit.not_found`	Rule, budget, or log entry not found.
500	`allternit.internal_error`	Internal gateway error.

Authentication — virtual-key management
Proxy — gateway middleware and routing
Security: DLP — secret scanning and injection detection
Security: Inference Hooks — pre/post-inference hooks

Get Started

Core Concepts

Gizzi Code CLI

Tools & ACI

Security & Enterprise

Admin API

BYOC Runtime

Surfaces

CI CD

Guides

Cookbooks

Release Notes

Overview

Base URL

Endpoints

Authentication

Usage

Query parameters

Example request

Example response

Logs

Query parameters

Example request

Example response

Routing decisions

Query parameters

Example request

Example response

Routing policies

List policies

Update a policy

Budgets

List budgets

Update a budget

Inference hooks

DLP rules

Error codes

​Overview

​Base URL

​Endpoints

​Authentication

​Usage

​Query parameters

​Example request

​Example response

​Logs

​Query parameters

​Example request

​Example response

​Routing decisions

​Query parameters

​Example request

​Example response

​Routing policies

​List policies

​Update a policy

​Budgets

​List budgets

​Update a budget

​Inference hooks

​DLP rules

​Error codes

​Related pages

Overview

Base URL

Endpoints

Authentication

Usage

Query parameters

Example request

Example response

Logs

Query parameters

Example request

Example response

Routing decisions

Query parameters

Example request

Example response

Routing policies

List policies

Update a policy

Budgets

List budgets

Update a budget

Inference hooks

DLP rules

Error codes

Related pages