API Reference

Complete reference for the AisaysAI API. Base URL: https://api.aisays.ai

Responses

POST/v1/chat/completionsAuth

Create a chat completion (OpenAI format)

POST/v1/responsesAuth

Create a response (OpenAI Responses API — native Codex CLI protocol)

POST/anthropic/v1/messagesAuth

Create a message (Anthropic format)

Quick Example

// Create a chat completion
const response = await fetch("https://api.aisays.ai/v1/chat/completions", {
  method: "POST",
  headers: {
    "Content-Type": "application/json",
    "Authorization": "Bearer YOUR_API_KEY",
  },
  body: JSON.stringify({
    model: "deepseek/deepseek-v4-flash",
    messages: [{ role: "user", content: "Hello!" }],
  }),
})

const data = await response.json()
console.log(data.choices[0].message.content)

Authentication

API endpoints marked with Auth require authentication via a Bearer token in the Authorization header.

You can create API keys from the API Keys page in your workspace.

Authorization: Bearer sk-or-xxxxxxxxxxxxxxxxxxxxxxxx

Model identity & verification

Models don't know who they are from their weights alone — a model's name and version are supplied at inference via a system prompt, not baked into its parameters. If you ask a model "what model are you?" through a gateway, it may misidentify itself (an industry-wide limitation known as identity confusion). This is not a routing problem: Aisays.ai always routes your request to the exact model you select.

How to verify the model you received

Trust the model field in the API response — it is set by Aisays.ai and reflects the exact model that served your request — not the model's self-reported text:

{
  "model": "deepseek/deepseek-v4-flash",
  "choices": [ { "message": { "content": "..." } } ]
}

Making a model state a specific identity

Send your own system message and the model will follow it:

{ "role": "system", "content": "You are DeepSeek V4 Flash." }

Our Chat playground auto-injects a model-identity system message by default (toggle it with the fingerprint icon) so models self-identify correctly. The raw /v1 API never injects model-identity prompts — it forwards your messages as sent (technical fields like the model ID are normalized server-side for routing and billing).

// Create a chat completion const response = await fetch("https://api.aisays.ai/v1/chat/completions", { method: "POST", headers: { "Content-Type": "application/json", "Authorization": "Bearer YOUR_API_KEY", }, body: JSON.stringify({ model: "deepseek/deepseek-v4-flash", messages: [{ role: "user", content: "Hello!" }], }), }) const data = await response.json() console.log(data.choices[0].message.content)