OpenAI-compatible API

Chat Completions and Responses against the Asale gateway, with the official OpenAI SDKs. Change the base URL and the key; the rest of your code is untouched.

The OpenAI dialect is the broadest one the gateway speaks, and the one most third-party libraries already target. If your code uses the openai package — or anything that imitates it — this is the page you want.

Read the overview first for keys, models, billing and errors; everything here is the OpenAI-specific part.

Connection


Base URL	`https://gw.asale.ai/v1`
Auth header	`Authorization: Bearer sk-asale-...`
SDK env var	`OPENAI_API_KEY`, `OPENAI_BASE_URL`

The /v1 belongs to the base URL here: OpenAI's clients append /chat/completions to whatever you give them.

Endpoints

Method	Path	Notes
POST	`/v1/chat/completions`	The main one.
POST	`/v1/responses`	The Responses API. Codex ≥ 0.146 speaks only this.
POST	`/v1/completions`	Legacy text completions, mapped onto the same path.
GET	`/v1/models`	What the platform will match right now.

Call it

curl https://gw.asale.ai/v1/chat/completions \
  -H "Authorization: Bearer $ASALE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gpt-5",
    "messages": [{"role": "user", "content": "Hello"}],
    "stream": false
  }'

# pip install openai
import os
from openai import OpenAI

client = OpenAI(
    api_key=os.environ["ASALE_API_KEY"],
    base_url="https://gw.asale.ai/v1",
)

resp = client.chat.completions.create(
    model="gpt-5",
    messages=[{"role": "user", "content": "Hello"}],
)
print(resp.choices[0].message.content)

// npm i openai
import OpenAI from "openai";

const client = new OpenAI({
  apiKey: process.env.ASALE_API_KEY,
  baseURL: "https://gw.asale.ai/v1",
});

const resp = await client.chat.completions.create({
  model: "gpt-5",
  messages: [{ role: "user", content: "Hello" }],
});
console.log(resp.choices[0].message.content);

Streaming

stream: true returns text/event-stream with chat.completion.chunk events and a final data: [DONE], exactly as the OpenAI API does. Your existing parser needs no changes.

stream = client.chat.completions.create(
    model="gpt-5",
    messages=[{"role": "user", "content": "Count to five"}],
    stream=True,
)
for chunk in stream:
    delta = chunk.choices[0].delta.content
    if delta:
        print(delta, end="", flush=True)

Tool calling

tools and tool_choice pass through, and the assistant's tool_calls come back in the same shape. Send the result as a role: "tool" message with its tool_call_id, as usual.

The model behind a request may be served by a publisher on any supported vendor, and the gateway translates between wire formats. Provider-specific opaque blobs — a reasoning signature, an encrypted_content — are only handed back when the request and the server speak the same wire; otherwise they are dropped rather than forged.

Practical notes

Set max_tokens. It is what the balance hold is computed from. Leaving it out makes the hold conservative, which can read as "insufficient balance" on an account that has plenty for the actual reply.
Model names are the platform's. GET /v1/models is the list; the market shows which of them have supply.
Images work where the model does. Request bodies are capped at 16 MiB, which is well above a typical inlined image; past that the gateway answers 413.
n is ignored — one completion per request.