AI for govtech in India

Government departments, PSUs and civic-tech vendors in India increasingly draft citizen replies, summarise grievances and translate notices with the help of a language model — and the honest answer about where that data goes matters more here than almost anywhere else. unoblox provides one rupee-billed, OpenAI-compatible endpoint for GPT, Claude, DeepSeek, Qwen, Llama and more, invoiced monthly in rupees with GST, rather than a foreign-currency subscription.

What govtech teams typically draft with it

Realistic uses include summarising an RTI request or grievance for triage, drafting a first-pass reply that an officer reviews before it's sent, translating a public notice into another Indian language, and turning a long circular or policy document into a shorter internal summary. Every one of these keeps a human decision-maker in the loop; unoblox has no workflow or approval logic of its own.

Matching a model to sensitivity and volume

Model	₹ / 1M input	₹ / 1M output	Typical use
qwen/qwen3-1.7b	0 (free)	0 (free)	Free pilot for an internal drafting tool
deepseek-ai/deepseek-v4-flash	9.07	18.14	High-volume grievance triage summaries
qwen/qwen3-235b-a22b-instruct-2507	9.07	55.44	Multilingual notice translation
openai/gpt-5	126	1008	Higher-stakes drafting reviewed by an officer

Every model's live rate is on /models; treat the figures above as indicative.

Data residency — the honest answer

This is the section that matters most for government use, so it's stated plainly: only the free, unoblox-hosted qwen/qwen3-1.7b model runs on infrastructure that unoblox controls. Every other model — the GPT, Claude, DeepSeek, larger Qwen, Llama, Gemma and Kimi lines — runs on its original provider's own infrastructure, which is not inside India. unoblox does not claim data residency for those models, and a department should classify its data and make its own residency call before routing anything sensitive to an externally hosted model. What unoblox does add for all of them is rupee billing, one GST invoice, one endpoint and no international card.

A sample call

from openai import OpenAI

client = OpenAI(
    base_url="https://api.unoblox.ai/v1",
    api_key="ub-gw-xxxxxxxxxxxxxxxxxxxx",
)

reply = client.chat.completions.create(
    model="qwen/qwen3-1.7b",
    messages=[{"role": "user", "content": "Draft a short acknowledgement reply to this grievance."}],
)

Frequently asked questions

Is citizen data stored on servers located in India? Only if you use the free, unoblox-hosted qwen/qwen3-1.7b model. Other models run on their original providers' infrastructure outside India — assess this against your department's data classification before use.

Is unoblox empanelled on a specific government procurement list? This page doesn't cover procurement empanelment status; contact the unoblox team directly for current documentation.

Can it make a decision on a grievance or application? No. It drafts and summarises; the decision and sign-off stay with your officers.

Can it translate a notice into multiple Indian languages? Yes, using a general multilingual model, reviewed by a language-competent officer before publishing.

Is there a free tier to pilot before procurement? Yes, qwen/qwen3-1.7b is free to call.

How is usage invoiced for a government department? One monthly GST invoice in rupees covering all usage, the same as any other unoblox customer.

Get started in rupees → https://unoblox.ai/sign-in