Skip to main content

​
Getting started

The full OpenAPI description that powers this reference lives alongside the docs. You can import it into Postman, Hoppscotch, or your favourite SDK generator to bootstrap a client automatically.

AI Stats Gateway OpenAPI

View the specification file
 All production traffic should target https://api.ai-stats.phaseo.app.

​
Authentication

All endpoints require a gateway API key to help mitigate abuse. Supply it as an Authorization: Bearer header. Keys have the form aistats_v1_k_ and map back to the teams you manage inside the dashboard. 
Authorization: Bearer aistats_v1_k_XXXX_example
If authentication fails you will receive a 401 error response.

​
Endpoint overview

The table below outlines the main endpoints available in the AI Stats Gateway right now:
EndpointDescription
POST /v1/chat/completionsRoute chat completions to the healthiest configured provider, with optional streaming via SSE.
POST /v1/moderationScore text input against provider moderation policies.
POST /v1/embeddingsGenerate embeddings for text input.
GET /v1/generation/{id}Retrieve the stored audit record for a previous request.
GET /v1/healthInspect provider health, routing scores, and breaker status.
GET /v1/modelsList available models across all providers along with their capabilities.
Use the include_endpoints query parameter on GET /v1/models to reproduce the per-surface catalogues (chat, moderations, embeddings, images, video, or audio) without bouncing between separate endpoints. With the table below showing the upcoming endpoints planned for future releases:
EndpointDescription
POST /v1/images/generationsGenerate one or more images from a prompt.
POST /v1/video/generationSubmit a video generation job to the upstream provider.
POST /v1/audioUse an audio model.
GET /v1/api-keysList and manage your API keys.
Responses include an optional usage and meta object with your usage, in both tokens and pricing, broken down into clear lines to see what costs what, as well as detailed timing and performance metadata, such as throughput, latency, and generation time, to support auditing and debugging.

​
Rate limits & billing

We apply apply no rate limits directly on our platform, any rate limits are upstream, applied by the providers themselves, we are working to get higher rate limits from the upstream providers to ensure that our users have a smooth experience. We apply credit checks prior to all requests made, ensuring that you have a minimum of $1.00 of credit in the team wallet. We do this to ensure that users do not accidentally rack up large bills on their accounts. You can top up your team wallet via the dashboard at any time, with payments typically going through within a few minutes.