PUBLIC DEVELOPER GUIDE

One endpoint.
Your next integration.

Use sri's OpenAI-compatible chat endpoint from Python, JavaScript or cURL. You can read this guide without an account.

1. Create a key

Create a free account, then open API keys in your workspace. Keep keys on your server or in environment variables. Never put a private API key in frontend code or public repositories.

2. Choose a model

Find the exact API ID in the model catalog. Free accounts can call the models marked Free without an extension. Full access provides additional models; DeepSeek and Qwen require your own connected provider account. See setup.

3. Send a request

Python · pip install openai

import os
from openai import OpenAI

client = OpenAI(
    base_url="https://srisai.app/v1",
    api_key=os.environ["SRI_API_KEY"],
)

reply = client.chat.completions.create(
    model="gpt-6-luna",
    messages=[{"role": "user", "content": "Hello!"}],
)
print(reply.choices[0].message.content)

Streaming

Add stream=True and read the deltas as they arrive. Endpoint: POST /v1/chat/completions. Authentication: Authorization: Bearer YOUR_API_KEY.

stream = client.chat.completions.create(
    model="gpt-6-luna",
    messages=[{"role": "user", "content": "Explain recursion."}],
    stream=True,
)
for chunk in stream:
    if chunk.choices:
        print(chunk.choices[0].delta.content or "", end="")

Limits and usage

Free: up to 50 requests daily. Full access: up to 2,000 requests daily. Chat and API share one allowance per account. The allowance resets at 00:00 UTC (05:30 IST). Failed responses are refunded; an interrupted stream without a completion is also refunded. Requests must still obey provider availability and rate limits.

Errors and recovery

400: Check the model ID and request payload.
401: Check your API key and account access.
403: Check your plan and model permissions.
409: A provider connection may need syncing. Open the connection page.
429: Check daily usage or wait before retrying a rate-limited request.
5xx or a streaming error: Retry with a bounded delay or choose another available model. Do not retry indefinitely.

Compatibility and model identity

The API supports OpenAI-format chat requests; it is not a guarantee of every OpenAI API feature. Supported options differ by model. The catalog shows sri's configured display labels and route IDs. A supported fallback may use another model. Inspect returned model metadata where available.

Data and support

Requests are sent to the selected provider, or a backup when used. See privacy for storage and connection details. Check application status or contact support with a redacted error and the model ID. Never send your API key.

One endpoint.Your next integration.