Gradium Voice Launches on AWS as a SaaS Subscription and a SageMaker Model Image
Gradium is now available as a managed SaaS subscription on AWS Marketplace, billable through any AWS account and counted toward existing committed spend. For teams that need inference inside their own VPC, the same TTS model also ships as a deployable Amazon SageMaker image with streaming inference. This post covers both deployment paths, what's included, and which one fits which constraint.
AWS Marketplace: managed SaaS subscription
The Gradium AWS Marketplace listing gives access to the full Gradium voice AI platform without requiring a separate vendor contract or billing relationship. Subscriptions are billed directly through your AWS account, and usage counts toward any existing AWS committed spend agreement.
The L plan includes:
- 45 million credits per month, approximately 1,000 hours of TTS or 4,100 hours of STT
- Real-time TTS and STT over streaming WebSocket APIs
- Instant voice cloning from a short audio sample
- 15 concurrent connections
- Five supported languages: English, French, German, Spanish, and Portuguese
No infrastructure provisioning is required. Gradium operates the platform entirely on AWS, with no GPUs to configure and no self-hosted models to maintain.
Amazon SageMaker: in-VPC deployment
For organizations with data residency requirements or compliance policies that prevent audio from leaving a private AWS environment, Gradium's TTS model is available as a deployable SageMaker endpoint. Inference runs over SageMaker's streaming API, so the low time-to-first-audio characteristics of the managed service are preserved when the model runs in your own account.
This option fits teams operating under HIPAA or internal data-handling policies that prohibit sending audio to a third-party endpoint, as well as teams that already maintain SageMaker infrastructure and prefer to keep inference inside their existing operational footprint.
Choosing between the two
The decision usually reduces to one constraint: where the audio is allowed to be processed.
If audio can leave your AWS account, the Marketplace SaaS plan is the lower-overhead path. Gradium handles capacity, scaling, model updates, and GPU provisioning. You consume the API.
If audio must stay inside your VPC, the SageMaker image gives you that boundary at the cost of managing the endpoint lifecycle, autoscaling, and instance selection yourself.
Frequently asked questions
Does a Gradium Marketplace subscription count toward my AWS committed spend?
Yes. Usage is billed through your AWS account.
What languages are supported?
The full set of languages supported by Gradium, for both TTS and STT.
Do I need to manage any infrastructure for the SaaS plan?
No. The Marketplace subscription is fully managed by Gradium. No instances, containers, or GPU resources to configure.
What is the difference between the Marketplace plan and the SageMaker image?
The Marketplace plan is a fully managed, multi-tenant SaaS offering. The SageMaker image deploys Gradium's TTS model directly into your own AWS account and VPC, giving you full control over data boundaries at the cost of managing the endpoint lifecycle yourself. The Gradium SageMaker image is available here.
How much does the Gradium L plan cost?
Full pricing is available on the Gradium AWS Marketplace listing.