Deepgram's India Endpoint Is Now Generally Available

Today, the Deepgram India endpoint is generally available to every customer. Any organization with Indian data-residency requirements, whether based in India or serving Indian users from abroad, can now run speech-to-text, text-to-speech, and Voice Agent workloads on Deepgram, with storage and inference inside India, using the same API, models, and SDKs available on our Global, EU, and Australia endpoints.

India has been among the most-requested regions from our customers, with demand concentrated in banking, payments, insurance, and the software vendors that serve them. This launch is a direct response to that demand.

For regulated industries in India, where voice data is stored and processed is a procurement requirement rather than a preference. Data protection law applies broadly, and the binding localization obligations sit in sector regulation covering payments and insurance. Teams in those sectors consistently tell us they go further than the rules strictly require, because auditors scrutinize any customer data that leaves the country. For them, keeping audio and transcripts in country is what separates a deployment that passes review from one that stalls.

Until now, teams building on Deepgram in India could route audio to our Global or regional endpoints, which introduced cross-border exposure, or deploy Deepgram self-hosted within their own Indian infrastructure, which requires the engineering capacity to operate it in production. The India endpoint provides a fully managed, fully onshore option that uses the same Deepgram Voice AI API as our Global, EU, or Australia endpoints.

The India endpoint is production-ready and available to all customers, with no waitlist and no enterprise-only restriction. It runs in AWS's ap-south-2 region (Hyderabad), and pricing matches our Global, EU, and Australia rates at launch.

Adoption requires a single change. Point your integration at api.in.deepgram.com, and your existing API keys and SDK integrations continue to work. There is no separate account, no migration project, and no change to how you call the API.

By default, storage and inference for your audio, transcripts, and speech output happen in India. Opting out of model improvement (mip_opt_out=true) extends that to all processing and access.

The residency commitment is specific. For speech-to-text and text-to-speech on Deepgram models, storage and inference happen on Indian infrastructure (AWS ap-south-2, Hyderabad). Two boundaries are explicit, so compliance teams can review the full picture:

• Voice Agent LLM. If a Voice Agent uses a managed third-party LLM (the reasoning step), that provider processes the request outside India. Confirm the provider's residency and processing guarantees independently; Deepgram runs listen and speak in-country.
• Operational metadata and billing are processed in the US.

For the strictest requirements, where regulated entities avoid any customer data leaving the country, opt out of model improvement (mip_opt_out=true). That keeps customer content out of any training workflow, so no content is accessed or processed across Indian borders for any purpose.

The India endpoint provides the same API surface as our Global, EU, and Australia regions. Speech-to-text is available at /v1/listen and /v2/listen, text-to-speech at /v1/speak and /v2/speak, Voice Agent at /v1/agent/converse, and text intelligence at /v1/read.

Deepgram model availability is at parity with our EU and Australia regions, so the speech-to-text and text-to-speech models teams already run on Deepgram are available in India without changes.

Running Voice Agent on the India endpoint keeps Deepgram's speech-to-text and text-to-speech onshore. The agent's LLM step runs on the model provider you select or provide, and that provider determines where the reasoning is processed. Confirm the provider's residency and processing guarantees independently. Teams with the strictest requirements typically bring their own model, hosted in India, and point the agent at it.

Set the base URL to the India endpoint in your existing client. The rest of your code stays the same.

from deepgram import DeepgramClient, ClientOptionsFromEnv

# Point your existing integration at the India endpoint
deepgram = DeepgramClient(
    "YOUR_API_KEY",
    ClientOptionsFromEnv(url="https://api.in.deepgram.com"),
)

Voice Agent connections use api.in.deepgram.com/v1/agent/converse. All customer payloads route through Indian infrastructure.

Indian teams evaluating voice AI have real in-country options, and the differences between them are worth understanding. Some global providers offer India as a storage region while inference continues to run elsewhere, so processing still crosses borders. Others make an in-country deployment available only through infrastructure the customer deploys and operates. Deepgram runs storage and inference in India as a fully managed service, with no infrastructure for the customer to run, and without asking teams to trade model quality for a local footprint.

That distinction matters most in banking, payments, insurance, and the vendors building for them, where an onshore guarantee separates a deployment that passes review from one that stalls. Deepgram holds SOC 2 Type I and Type II certifications, and the India endpoint extends that compliance posture with in-country processing and storage.

The India endpoint is live now and available to every Deepgram customer.

• Read how regional endpoints work →
• India endpoint quickstart →

To move a workload to India, update your base URL to api.in.deepgram.com, or contact your account team to plan the transition.

The India endpoint is part of a broader expansion. Over the past year, Deepgram has extended where voice AI runs, from Dedicated deployments to our EU endpoint, then Australia, and now India, while continuing to expand the languages our models support. The goal is consistent: Deepgram should run where your business runs, in the language your customers speak. India is the next step, and more regions will follow.