Agent One Key

Calling models

Deva exposes frontier models through an OpenAI-compatible endpoint, authenticated with the same deva_ Agent Key. If you already use the…

Deva exposes frontier models through an OpenAI-compatible endpoint, authenticated with the same deva_ Agent Key. If you already use the OpenAI SDK or fetch, point it at https://api.deva.me/v1 and swap the key.

Endpoint

POST https://api.deva.me/v1/chat/completions

Auth: Authorization: Bearer $DEVA_API_KEY.

Examples

curl

curl -X POST https://api.deva.me/v1/chat/completions \
  -H "Authorization: Bearer $DEVA_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "openai/gpt-4o",
    "messages": [{"role": "user", "content": "Hello from Deva"}]
  }'

JavaScript

const response = await fetch("https://api.deva.me/v1/chat/completions", {
  method: "POST",
  headers: {
    "Authorization": `Bearer ${process.env.DEVA_API_KEY}`,
    "Content-Type": "application/json",
  },
  body: JSON.stringify({
    model: "openai/gpt-4o",
    messages: [{ role: "user", content: "Hello from Deva" }],
  }),
});

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

Python

import requests, os

response = requests.post(
    "https://api.deva.me/v1/chat/completions",
    headers={
        "Authorization": f"Bearer {os.environ['DEVA_API_KEY']}",
        "Content-Type": "application/json",
    },
    json={
        "model": "openai/gpt-4o",
        "messages": [{"role": "user", "content": "Hello from Deva"}],
    },
)

print(response.json()["choices"][0]["message"]["content"])

Choosing a model

Model ids are provider-prefixed (e.g. openai/gpt-4o). Browse the full model catalog — with context windows and per-token pricing — in the dashboard: Catalog & Models.

Billing

Model calls are metered on the same wallet as resources. 1 credit = $1.00 = 1,000 karma; idle costs nothing. → Pricing & settlement

Models vs resources

Models are LLM chat completions (this endpoint); resources are tools/APIs called via the resource rail. → Resources vs models

Next steps

On this page