Migrate from OpenAI to India-Hosted API in One Line

The Migration

If you use the OpenAI SDK today, migrating to India-hosted inference takes exactly one line change. No code rewrite. No new SDK. No new abstractions.

IndicStack's API is OpenAI-compatible - same endpoint format, same request/response schema, same streaming behavior. You change your base_url and you're done.

---

Python

from openai import OpenAI

# Before - routed through US servers
# client = OpenAI(
#     base_url="https://api.openai.com/v1",
#     api_key="sk-..."
# )

# After - India-hosted inference
client = OpenAI(
    base_url="https://api.indicstack.ai/v1",
    api_key="isk_your_api_key"
)

# Everything below stays identical
response = client.chat.completions.create(
    model="sarvam-30b",
    messages=[
        {"role": "system", "content": "You are a helpful assistant."},
        {"role": "user", "content": "Explain the DPDP Act in simple terms"}
    ],
    temperature=0.7,
    max_tokens=1024
)

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

---

Node.js / TypeScript

import OpenAI from 'openai';

// Before
// const client = new OpenAI({ baseURL: 'https://api.openai.com/v1' });

// After - one line
const client = new OpenAI({
  baseURL: 'https://api.indicstack.ai/v1',
  apiKey: 'isk_your_api_key',
});

const response = await client.chat.completions.create({
  model: 'sarvam-30b',
  messages: [
    { role: 'user', content: 'Explain GST input tax credit' }
  ],
  stream: true,
});

for await (const chunk of response) {
  process.stdout.write(chunk.choices[0]?.delta?.content || '');
}

---

cURL

curl https://api.indicstack.ai/v1/chat/completions \
  -H "Authorization: Bearer isk_your_api_key" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "sarvam-30b",
    "messages": [
      {"role": "user", "content": "Explain DPDP Act in simple terms"}
    ],
    "temperature": 0.7
  }'

---

What Stays the Same

Everything you already know works unchanged:

• Chat completions - same /v1/chat/completions endpoint
• Streaming - SSE with same stream: true parameter
• Function calling - same tools and tool_choice parameters
• JSON mode - same response_format: { type: "json_object" }
• System prompts - same system role messages
• Temperature, max_tokens, top_p - all standard parameters supported
• Error codes - same HTTP status codes and error shapes

---

What Changes

Dimension	OpenAI	IndicStack
Base URL	api.openai.com/v1	api.indicstack.ai/v1
API key prefix	sk-	isk_
Models	GPT-4o, GPT-4o-mini, etc.	sarvam-30b, krutrim-2, etc.
Inference location	US (Virginia/Oregon)	India
Billing	USD	INR + GST
Data jurisdiction	US law (CLOUD Act)	Indian law

---

Choosing a Model

IndicStack offers models across three tiers:

Economy - Fast, cheap, high-volume. Best for chatbots, classification, simple extraction.

• krutrim-2 (12B, 22 Indic languages)
• openhathi-7b (Hindi-focused)

Default - Balanced performance and cost. Best for most production workloads.

• sarvam-30b (30B MoE, 22+ languages, Apache 2.0)
• sarvam-m (24B, strong multilingual)

Premium - Highest quality for complex reasoning.

• sarvam-105b (106B, complex tasks)

All models support Indian languages natively - not through a translation layer.

---

Streaming

Streaming works identically to OpenAI. Server-sent events, same chunk format:

stream = client.chat.completions.create(
    model="sarvam-30b",
    messages=[{"role": "user", "content": "Write a poem in Tamil"}],
    stream=True
)

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

---

Environment Variables

For production deployments, set your base URL and key via environment:

export OPENAI_BASE_URL=https://api.indicstack.ai/v1
export OPENAI_API_KEY=isk_your_api_key

Most OpenAI SDK wrappers (LangChain, LlamaIndex, Vercel AI SDK) respect these environment variables, so you may not need to change any code at all.

---

LangChain

from langchain_openai import ChatOpenAI

llm = ChatOpenAI(
    base_url="https://api.indicstack.ai/v1",
    api_key="isk_your_api_key",
    model="sarvam-30b"
)

response = llm.invoke("Summarize RBI's data localization circular")

---

Vercel AI SDK

import { openai } from '@ai-sdk/openai';
import { generateText } from 'ai';

const { text } = await generateText({
  model: openai('sarvam-30b', {
    baseURL: 'https://api.indicstack.ai/v1',
    apiKey: 'isk_your_api_key',
  }),
  prompt: 'Explain DPDP Act compliance requirements',
});

---

What You Get After Migration

• India data residency - All inference on Indian infrastructure
• INR billing - Predictable costs, proper GST invoices
• No CLOUD Act exposure - Indian company, Indian jurisdiction
• Per-client metering - Track usage by project or customer
• No-training guarantee - Your data is never used for model improvement
• Configurable retention - 30 days, 7 days, or zero

---

Next Steps

1. Request early access to get your API key
2. Change your base URL
3. Run your existing test suite - everything should pass
4. Deploy to production on Indian infrastructure