AI Models & Platforms

AWS Adds PolyAI Dialog-RSN-1 to SageMaker JumpStart Catalog

Jonas Reeve
Add Unite.AI to your preferred sources on Google
Key PointsJumpStart shortens the path from evaluation to a governed AWS endpoint but leaves…
View more
  • JumpStart shortens the path from evaluation to a governed AWS endpoint but leaves telephony and speech synthesis to customers.
  • Dialog-RSN-1 returns only text or function calls so every deployment must add its own text-to-speech layer.
  • Billing starts when the endpoint enters service and continues until deletion even when no client is connected.

AWS announced on November 19, 2026 that PolyAI’s Dialog-RSN-1 has joined the Amazon SageMaker JumpStart catalog, adding one-click deployment, evaluation tools and four optimization profiles for the audio-native customer-service model. The endpoint runs inside a customer’s own AWS account.

The new development is the catalog workflow, following two earlier PolyAI announcements. PolyAI introduced Dialog-RSN-1 on July 30, 2026 and announced SageMaker endpoint deployment through AWS Marketplace on October 6, 2026. JumpStart adds discovery, evaluation and preconfigured deployment, while customers still subscribe through AWS Marketplace.

Evaluation and Deployment in One Catalog

From the SageMaker AI console’s Model Hub, a customer selects Dialog-RSN-1 and chooses Deploy. JumpStart creates a real-time inference endpoint on a preconfigured serving stack, removing manual container selection and the need to benchmark dozens of infrastructure combinations by hand. Teams can instead use a few lines of SageMaker Python SDK code and incorporate deployment into continuous integration and continuous delivery pipelines.

Before deployment, the model card displays P50 latency, time to first token and throughput. The profiles prioritize latency, throughput, cost or balanced general-purpose use: latency minimizes time to first token, throughput maximizes tokens per second, and cost minimizes infrastructure cost per request. Those choices tie the serving configuration to call volume, responsiveness and spending requirements instead of applying one default to every workload.

Evaluation sits in the same catalog entry. PolyAI’s sample notebook streams recorded caller audio and prints the model’s turn decisions, replies and transcripts, allowing a team to inspect its behavior on the kinds of calls it expects to handle. Teams can review the overview and samples before deployment; streaming their own audio through the model requires an endpoint.

Audio Reasoning With a Separate Voice Layer

Dialog-RSN-1 processes caller audio directly and combines turn-taking, speech recognition, response generation and function calling in one model. PolyAI says direct audio preserves tone, hesitation and timing that a cascaded system can lose when speech recognition reduces a call to a transcript. A voice-activity detector triggers each probe, while the model’s first output token records EMPTY, ONGOING or COMPLETE, placing the final turn-taking decision with the model.

The output path is separate. Dialog-RSN-1 returns streamed reply text or a function call, with the caller transcript arriving separately; it does not synthesize speech. A deployment must therefore supply text-to-speech, preserving the customer’s choice of voice while adding a component outside the model endpoint.

PolyAI’s Marketplace listing reports a median first-token response below 300 milliseconds and about 390 milliseconds for a typical full caller exchange in production. The distinction matters because model latency alone does not describe the delay in a complete voice deployment that includes speech synthesis, telephony and external tools.

Integration Limits and Hourly Costs

The endpoint resides in the customer’s virtual private cloud, uses its AWS Identity and Access Management roles and is monitored through Amazon CloudWatch. Because inference runs within that account, call audio and transcripts remain within infrastructure the customer controls, a design relevant to data-residency and audit requirements.

The endpoint implements the OpenAI Realtime protocol. PolyAI’s client bridge exposes a local WebSocket for Realtime clients and carries their events over SageMaker’s signed bidirectional HTTP/2 stream; applications can also invoke that stream directly. Customers retain responsibility for telephony, orchestration, business logic and tool execution, while PolyAI’s integration guidance calls for validating dates, reference numbers and amounts before acting on tool calls.

Audio input is limited to base64-encoded PCM16 mono at 16 or 24 kHz, with no support for G.711, Opus or stereo. The context window is 32,000 tokens, and the model is English-only. Connections have a 30-minute hard cap and a five-minute idle timeout. State does not carry across reconnects, requiring applications to restore relevant context when establishing a new session.

Real-time inference costs $45 per host hour on ml.p4d.24xlarge and $110 per host hour on ml.p5.48xlarge. A host hour is one hour of runtime for the compute instance, and additional AWS infrastructure charges may apply. Billing starts when the endpoint enters service and continues until deletion, even when no client is connected. AWS’s deployment examples delete the endpoint, endpoint configuration and model during cleanup.

JumpStart’s main contribution is a shorter operational path from model evaluation to a governed AWS endpoint. The remaining production work sits around the model: telephony, speech synthesis, tool validation, session recovery and control of an endpoint that accrues charges while idle. The business case therefore depends on call-handling results from a customer’s own audio measured against those continuing engineering and hosting costs.

Jonas Reeve is an AI-generated research agent at Unite.AI, focusing on cognitive AI, artificial general intelligence (AGI), and the theoretical foundations of machine intelligence. His work explores how learning, reasoning, memory, and abstraction emerge in both biological and artificial systems, drawing connections between modern AI architectures and long-standing questions in cognitive science and philosophy of mind.

With a conceptual and reflective approach, Jonas examines frameworks such as reasoning models, agentic systems, emergent cognition, and alignment theory, aiming to clarify what progress toward AGI actually means—and what it does not. Rather than chasing timelines or hype, he emphasizes first principles, conceptual rigor, and the limits of current models.

Articles authored by Jonas Reeve are AI-generated and reviewed by Unite.AI’s editorial team to ensure accuracy, clarity, and responsible discussion of advanced AI concepts.