An OpenAI-compatible API for many models

Keep the code you already have. Point the official OpenAI SDK, or any tool that speaks the OpenAI API, at ExosAI and switch between supported models by changing one string.

Switch in two lines

Set the base URL to https://api.exosai.tech/v1 and use your ExosAI API key. Everything else stays the same: chat completions, streaming, system prompts and message history.

from openai import OpenAI

client = OpenAI(base_url="https://api.exosai.tech/v1", api_key="YOUR_EXOSAI_KEY")
reply = client.chat.completions.create(
    model="MODEL_ID_FROM_THE_CATALOG",
    messages=[{"role": "user", "content": "Hello!"}],
)
print(reply.choices[0].message.content)

Node.js and curl

import OpenAI from "openai";

const client = new OpenAI({ baseURL: "https://api.exosai.tech/v1", apiKey: process.env.EXOSAI_API_KEY });
const reply = await client.chat.completions.create({
  model: "MODEL_ID_FROM_THE_CATALOG",
  messages: [{ role: "user", content: "Hello!" }],
});
console.log(reply.choices[0].message.content);

curl https://api.exosai.tech/v1/chat/completions \
  -H "Authorization: Bearer $EXOSAI_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model": "MODEL_ID_FROM_THE_CATALOG", "messages": [{"role": "user", "content": "Hello!"}]}'

What you get

• A live model catalog with exact model IDs and published prices per million tokens
• Scoped API keys: one per project, limited to the models you choose, revocable at any time
• Usage you can read: requests, tokens and charges in your dashboard
• A playground to compare models before you write code

Questions

More