Hermes Agent Bangladesh for Developers: OpenAI-Compatible API Guide

For developers building AI features into their applications, Hermes Agent Bangladesh provides an OpenAI-compatible API endpoint. This means you can use Hermes as a drop-in replacement for the OpenAI API — the same code, same SDK, same request format — while keeping your data in Bangladesh and avoiding USD billing.

What "OpenAI-Compatible" Means

Hermes Agent Bangladesh routes through LiteLLM, which implements the OpenAI API specification. Concretely, this means:

• The same /v1/chat/completions endpoint format
• The same JSON request structure (model, messages, temperature, etc.)
• The same JSON response structure (choices, message, usage)
• Compatibility with the official OpenAI Python SDK, JavaScript SDK, and any library that supports a custom base URL

The only change in your code is the base_url and api_key — everything else stays identical.

Python Integration

from openai import OpenAI

client = OpenAI(
    base_url="https://your-hermes-endpoint.snbdhost.com/v1",
    api_key="your-hermes-api-key"
)

response = client.chat.completions.create(
    model="hermes",
    messages=[
        {"role": "system", "content": "You are a helpful assistant for a Bangladesh business."},
        {"role": "user", "content": "What are your business hours?"}
    ],
    temperature=0.7
)

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

If your existing code uses openai.OpenAI() with the default base URL, you only need to add the base_url and update the api_key.

JavaScript / TypeScript Integration

import OpenAI from 'openai';

const client = new OpenAI({
  baseURL: 'https://your-hermes-endpoint.snbdhost.com/v1',
  apiKey: 'your-hermes-api-key',
});

const response = await client.chat.completions.create({
  model: 'hermes',
  messages: [
    { role: 'system', content: 'You are a customer service agent for a Dhaka e-commerce brand.' },
    { role: 'user', content: 'I want to return an item.' },
  ],
});

console.log(response.choices[0].message.content);

Using the Qdrant Memory API

Hermes Agent Bangladesh includes Qdrant vector memory. You can interact with it via the Qdrant HTTP API to:

• Upsert documents into your agent's knowledge base
• Search semantically similar content for RAG (retrieval-augmented generation) applications
• Delete or update documents as your knowledge base changes

Your Qdrant endpoint is provided in your Hermes dashboard credentials. The Qdrant Python and JavaScript clients connect to it using standard Qdrant API calls.

Streaming Responses

Hermes Agent Bangladesh supports streaming via the stream=True parameter — the same as OpenAI:

stream = client.chat.completions.create(
    model="hermes",
    messages=[{"role": "user", "content": "Explain BDIX peering in simple terms."}],
    stream=True
)

for chunk in stream:
    if chunk.choices[0].delta.content:
        print(chunk.choices[0].delta.content, end="")

Rate Limits and Token Usage

During the beta, the 10M token allowance is the primary constraint — there are no hard rate limits on requests per minute. Token usage is visible in your Hermes dashboard. The usage field in every API response shows tokens consumed per call.

Getting Your API Credentials

Apply for the Hermes Agent Bangladesh beta. Once approved, your dashboard includes:

• Your Hermes API base URL
• Your API key
• Your Qdrant endpoint and collection name
• Token usage remaining

The free beta includes 10 million tokens — more than enough for months of development and testing at typical application volumes.