How much does NVIDIA Nemotron (DeepInfra) speech-to-text cost per month for 1,000 minutes?

About $2.00 a month using $0.002/min (prices updated 2026-10-30). Free-tier credits are not subtracted.

Where does the NVIDIA Nemotron (DeepInfra) pricing on this page come from?

From the provider’s own public pricing page, last checked on 2026-10-30. The methodology page lists every source and rule.

NVIDIA Nemotron (DeepInfra) pricing

Nemotron 3.5 Transcribe is an open-weight speech model NVIDIA released in October 2026 under the OpenMDW-1.1 licence, which allows commercial use. It transcribes recorded audio in 40 languages, labels up to eight speakers and handles recordings of up to an hour. It does not do live streaming; NVIDIA's separate Nemotron ASR models cover that.

An open model is not a free service. NVIDIA does not sell this model by the minute, so the price ranked here is what DeepInfra, a third-party host, charges to run it. Downloading the weights costs nothing, but running them means paying for GPUs and the people who keep them running.

The model card has the hardware requirements and accuracy results.

Some “Sign up” links are affiliate links and may earn a commission. Rankings are computed from price only and never take affiliate status into account.

Every month I transcribe of audio.

Label who is speakingRun it on my own servers

NVIDIA Nemotron (DeepInfra) is the cheapest, at $2.00 a month. Google Cloud would cost 8.0 times as much.

Estimated monthly bill for 1,000 minutes, lowest first. 9 of 10 providers fit.

1NVIDIA Nemotron (DeepInfra)This pageNemotron 3.5 Transcribe, $0.002/min$2.00 per monthLowest priceVisit NVIDIA Nemotron (DeepInfra) (opens in a new tab)
Usage $2.00
Monthly total $2.00
Works out to $0.002 per minute.
Priced in the "Above 0" tier; every unit is billed at that tier's rate.
2AzureStandard batch transcription, $0.18/hr$3.00 per month+$1.00, 50.0% moreVisit Azure (opens in a new tab)
Usage $3.00
Monthly total $3.00
Works out to $0.003 per minute.
Priced in the "Above 0" tier; every unit is billed at that tier's rate.
3AssemblyAIUniversal-3.5 Pro (pre-recorded), $0.21/hr$3.50 per month+$1.50, 75.0% moreVisit AssemblyAI (opens in a new tab)
Usage $3.50
Monthly total $3.50
Works out to $0.0035 per minute.
Priced in the "Above 0" tier; every unit is billed at that tier's rate.
3Mistral VoxtralVoxtral Mini Transcribe 3, $0.0035/min$3.50 per month+$1.50, 75.0% moreVisit Voxtral (opens in a new tab)
Usage $3.50
Monthly total $3.50
Works out to $0.0035 per minute.
Priced in the "Above 0" tier; every unit is billed at that tier's rate.
5ElevenLabsScribe v2 (and v2 Medical), $0.22/hr$3.67 per month+$1.67, 83.3% moreVisit ElevenLabs (opens in a new tab)
Usage $3.67
Monthly total $3.67
Works out to $0.00367 per minute.
Priced in the "Above 0" tier; every unit is billed at that tier's rate.
6DeepgramNova-3 (pre-recorded), $0.0043/min$4.30 per month+$2.30, 115.0% moreVisit Deepgram (opens in a new tab)
Usage $4.30
Monthly total $4.30
Works out to $0.0043 per minute.
Priced in the "Above 0" tier; every unit is billed at that tier's rate.
7OpenAIGPT Transcribe, $0.0045/min$4.50 per month+$2.50, 125.0% moreVisit OpenAI (opens in a new tab)
Usage $4.50
Monthly total $4.50
Works out to $0.0045 per minute.
Priced in the "Above 0" tier; every unit is billed at that tier's rate.
8SarvamSpeech to Text (batch), $0.00515/min$5.15 per month+$3.15, 157.7% moreEstimateVisit Sarvam (opens in a new tab)
Usage $5.15
Monthly total $5.15
Works out to $0.00515 per minute. Converted to USD.
Priced in the "Above 0" tier; every unit is billed at that tier's rate.
Converted at 1 INR = 0.010518 USD (2026-10-31).
9Google CloudSpeech-to-Text V2 Standard (incl. Chirp), $0.016/min$16.00 per month+$14.00, 700.0% moreVisit Google Cloud (opens in a new tab)
Usage $16.00
Monthly total $16.00
Works out to $0.016 per minute.
Tiers are graduated: each band of usage is billed at its own rate (reached: Up to 500000).

Usage	$2.00
Monthly total	$2.00

Usage	$3.00
Monthly total	$3.00

Usage	$3.50
Monthly total	$3.50

Usage	$3.50
Monthly total	$3.50

Usage	$3.67
Monthly total	$3.67

Usage	$4.30
Monthly total	$4.30

Usage	$4.50
Monthly total	$4.50

Usage	$5.15
Monthly total	$5.15

Usage	$16.00
Monthly total	$16.00

Not available for this setup

RimeNot in this comparison.

Prices updated 2026-10-30.
Compared on transcription of recorded files (pre-recorded or batch). Live streaming is billed differently and costs more at some providers.
Priced in the "Above 0" tier; every unit is billed at that tier's rate.
Free-tier credits are not subtracted.
How we calculate this

Some “Visit” links are affiliate links: the provider may pay us if you sign up. That never changes the order, which is set by price alone.

Pricing by model

Nemotron 3.5 Transcribe pricingNew

Speech to text. Open-weight offline transcription in 40 languages with speaker labels, for recordings up to one hour.

Monthly usage	Price per minute
Any volume	$0.002

DeepInfra’s hosted price, $0.002 per minute of audio, with no separate speaker-label charge on the page. This is a third party’s price: NVIDIA does not sell this model by the minute. The weights are free under OpenMDW-1.1, but running them needs GPUs, so the self-hosting cost is worked out separately on this page and never ranked. Offline only: recorded audio up to one hour, no live streaming. Read from the DeepInfra page on 2026-10-30.

Source: deepinfra.com

Self-hosting or the hosted API

The Nemotron 3.5 Transcribe weights are free to download, but they need a GPU. This compares the hosted rate above with renting a GPU around the clock (730 hours a month). It is not part of the ranking.

Audio minutes a monthGPU cost per hour (USD)

Hosted API	$200.00 a month
1 rented GPU, always on	$2,109.70 a month
Lower GPU rent than hosted from	1,054,850 minutes a month
GPU rent per audio minute at full load	$0.000117

An estimate that counts GPU rental only: no staff, storage, bandwidth or spare capacity. Throughput is NVIDIA's rough figure for batched English transcription on one H100 80 GB with reasoning off; other languages, long recordings and reasoning may be slower, and a real service will not run at full load.

Default GPU price: runpod.io (NVIDIA H100 80GB PCIe, Runpod Secure Cloud, on demand, 2026-10-30). Throughput: huggingface.co (about 410 seconds of audio per second).

Features

Feature	NVIDIA Nemotron (DeepInfra)
Speaker labels	Yes
Languages	40
Self-hosting	Yes
EU data residency	No
Realtime voice	No

Frequently asked questions

How much does NVIDIA Nemotron (DeepInfra) speech-to-text cost per month for 1,000 minutes?: About $2.00 a month using $0.002/min (prices updated 2026-10-30). Free-tier credits are not subtracted.
Where does the NVIDIA Nemotron (DeepInfra) pricing on this page come from?: From the provider’s own public pricing page, last checked on 2026-10-30. The methodology page lists every source and rule.