Skip to main content
AI Gateway provides a unified interface for AI providers with caching, rate limiting, analytics, and cost controls. Use the API to create gateways and manage AI traffic.

Overview

Access the AI Gateway API:

Gateways

Manage AI Gateway instances.

Create a gateway

Create a new AI Gateway.
string
required
Your Cloudflare account ID
string
required
Gateway identifier (used in API calls)
number
Cache time-to-live in seconds (null to disable caching)
boolean
required
Invalidate cache when gateway is updated
boolean
required
Enable request/response logging
number
Rate limit window in seconds (null to disable)
number
Maximum requests per interval (null to disable)
string
required
Rate limiting algorithm: ‘fixed’ or ‘sliding’
string
The gateway ID
string
Your account ID
string
ISO 8601 timestamp when the gateway was created
string
ISO 8601 timestamp when the gateway was last modified

List gateways

Retrieve all AI Gateways in your account.
string
required
Your Cloudflare account ID
Filter gateways by ID
number
Page number for pagination
number
Number of gateways per page

Get a gateway

Retrieve details about a specific gateway.
string
required
The gateway ID
string
required
Your Cloudflare account ID

Update a gateway

Update gateway configuration.
string
required
The gateway ID to update
string
required
Your Cloudflare account ID

Delete a gateway

Delete an AI Gateway.
string
required
The gateway ID to delete
string
required
Your Cloudflare account ID

Logs

Manage and query AI Gateway request logs.

List logs

Retrieve logs for a gateway.
string
required
The gateway ID
string
required
Your Cloudflare account ID
string
Start date for logs (ISO 8601)
string
End date for logs (ISO 8601)
number
Page number for pagination
number
Number of logs per page

Get a log entry

Retrieve a specific log entry.
string
The log entry ID
string
ISO 8601 timestamp of the request
string
AI provider (e.g., ‘openai’, ‘anthropic’)
string
Model used for the request
number
HTTP status code
boolean
Whether the response was served from cache
number
Request duration in milliseconds

Get log request

Retrieve the request payload for a log entry.

Get log response

Retrieve the response payload for a log entry.

Delete logs

Delete logs for a gateway.
string
Delete logs before this date (ISO 8601)

Datasets

Manage datasets for AI model evaluation.

Create a dataset

string
required
The gateway ID
string
required
Your Cloudflare account ID
string
required
Dataset name
array
required
Array of test cases with input and expected output

List datasets

Get a dataset

Update a dataset

Delete a dataset

Evaluations

Run model evaluations using datasets.

Create an evaluation

string
required
The gateway ID
string
required
Your Cloudflare account ID
string
required
Dataset to evaluate against
string
required
Evaluation metric type

List evaluations

Get an evaluation

string
Evaluation ID
string
Evaluation status: ‘pending’, ‘running’, or ‘complete’
number
Evaluation score (0-1)
array
Detailed results per test case

Using AI Gateway

With OpenAI SDK

With Workers AI

Best practices

  1. Caching: Enable caching for repeated queries to reduce costs and latency
  2. Rate limiting: Set appropriate rate limits to control costs
  3. Logging: Enable logging for debugging and analytics
  4. Monitoring: Regularly review logs and analytics for usage patterns
  5. Security: Use authentication to protect your gateway from unauthorized access
  6. Cost control: Set up alerts for spending thresholds