Nemotron-3-Diarization (NVIDIA)

View on DeepInfra

Nemotron 3 Diarization is NVIDIA’s open-weight model for identifying who speaks when in an audio recording. It distinguishes up to eight speakers and supports both streaming audio and offline processing.

Its specialization is speaker diarization: assigning speaker labels and timestamps to speech segments. It can be combined with speech recognition to produce transcripts that distinguish individual speakers.

Description source

Identity and capabilities

Status: active
Local execution: Not verified
Context window: Unknown
Maximum output: Unknown
Image input: Unknown
Reasoning: Unknown
Tool calling: Unknown
First seen: 2026-10-16 00:00:07Z UTC
Last verified: 1 min ago

Current offers - prices in USD · per audio minute

Prices by provider, service tier, context and offer route
	DeepInfra
Input (audio)	$0.0006

Active promotions

No officially confirmed promotion applies to this model.

Recent price changes

No price changes in the last 7 days.