Models / NVIDIA/ Parakeet RNNT 0.6b

Parakeet RNNT 0.6b

NVIDIA · nvidia/parakeet-rnnt-0.6b

Speech to textTranscribes a recording into words

Input: audio. Output: text.InputOutput
Type
Open weightsCreative Commons Attribution 4.0
Languages
1
Size
0.6B

Context measured in tokens

Our take

Written Aug 2, 2026

Parakeet RNNT is a tiny downloadable speech-to-text model from NVIDIA that processes audio faster than almost anything else we track. It is built for speed on clean English recordings, but its accuracy falls apart on harder material like meetings or accented speech.

Who should pick it

Pick this for batch transcription of clean, high-quality English audio where throughput matters most. Use it on edge hardware where model size is a hard constraint, or in English-only settings with read-aloud or financial-call audio. Skip it if you need non-English languages, messy or accented audio, or a hosted service.

The case for it

  • Extremely fast: over five thousand times real time on benchmark hardware, an hour of audio in under a second.
  • Tiny at 0.6 billion parameters, making it easy to deploy on constrained hardware.
  • Strong on clean read-aloud audio at 1.3% word error rate, and 2.9% on financial calls.
  • Permissive Creative Commons Attribution 4.0 licence allows commercial use with attribution.

The case against it

  • Accuracy collapses on challenging audio: 14.3% word error rate on recorded meetings, more than eleven times worse than its clean read-aloud score.
  • English only — one language supported, with no measurements we hold for anything else.
  • No commercial hosting available; you must self-host.
00

How good is it?

TranscriptionTurning speech into text2 of 5Open ASR WER · 56th of 74

Words it gets right

93.4%

Misses roughly one word in 15, averaged over nine English test sets.

How fast it listens

5,407×5th of 62

an hour of audio in under a second, on the board's own hardware. Your machine will differ.

Languages

1

Listed on the model card. The accuracy above is English only.

Where it struggles
Read aloudaudiobooks, clean recording1.3%
Podcasts and videoeveryday internet audio8.5%
Accented speechspeakers from many countries13.6%
Meetingsa room, several people, far microphone14.3%

Percentage of words wrong on each set, lower better. Bars are scaled to this model's own worst case, not to the board.

The figures above come from the Open ASR Leaderboard, an independent public test that runs every model on the same recordings. It is the only measurement of transcription quality we know of, so there are no other scores to show.

Also scored, on boards we give no mark for
Harder read speech 22nd of 74Clean read speech 26th of 74European-accented speech 28th of 74Podcasts and video 43rd of 74Financial calls 51st of 74Accented speech 66th of 74Recorded meetings 66th of 74

These tests check whether a model follows instructions — a precondition for all the work above, but not a measure of how well that work is done, which is why they get no rating.

Every published score for this model9 scoresEvery figure we hold, from 9 boards, with who ran it and a link to the source — including the boards no rating above is built on.
5406.7independentsource ↗
6.7independentsource ↗
13.6independentsource ↗
2.9independentsource ↗
14.3independentsource ↗
8.5independentsource ↗
1.3independentsource ↗
2.6independentsource ↗
01

Can you run it yourself?

Fits in memory
weights load entirely on the card
Spills to system RAM
some weights offload; much slower
Too large
will not load even with offload
est
size is calculated; the verdict could change by 10%

Comfortable fit

A card many people ownFits in memory

GeForce RTX 4090 · 24 GB

Weights at Q4_K_M0.4 / 24 GBest
Spare memory21.2 GB spare
Usable context262Kwhat the spare memory holds; no published limit on record
Decode speed2217 tok/sest

Room to spare. 21.2 GB spare means a 10% error in the size would not change the answer.

One step upFits in memory

Apple M1 Pro (16-core GPU) · 32 GB

Weights at Q4_K_M0.4 / 32 GBest
Spare memory22.4 GB spare
Usable context262Kwhat the spare memory holds; no published limit on record
Decode speed385 tok/sest

Room to spare. 22.4 GB spare means a 10% error in the size would not change the answer.

On a MacFits in memory

Apple M1 (8-core GPU, 8GB unified) · 8 GB

Weights at Q4_K_M0.4 / 8 GBest
Spare memory4.4 GB spare
Usable context262Kwhat the spare memory holds; no published limit on record
Decode speed131 tok/sest

Room to spare. 4.4 GB spare means a 10% error in the size would not change the answer.

Your hardware
Checking your profile…

Memory use by level

Against a 24 GB card.

Q4_K_M
recommended
0.4 GBest
Fits in memory
Q5_K_M
0.4 GBest
Fits in memory
Q8_0
0.7 GBest
Fits in memory

All 3 sizes here are calculated, not measured. We hold no measured file for this model, so each size comes from the parameter count and every verdict above inherits that uncertainty.

Check against your own machine → · All 71 devices, with every size →

02

Models people weigh against Parakeet RNNT 0.6b

03

When we formed this view

Dates behind this page

Aug 2, 2026BenchmarkScored 13.6 on Accented speechleaderboard
Aug 2, 2026BenchmarkScored 2.9 on Financial callsleaderboard
Aug 2, 2026BenchmarkScored 14.3 on Recorded meetingsleaderboard
Aug 2, 2026BenchmarkScored 3.4 on European-accented speechleaderboard
Aug 2, 2026BenchmarkScored 8.5 on Podcasts and videoleaderboard
Aug 2, 2026BenchmarkScored 1.3 on Clean read speechleaderboard
Aug 2, 2026BenchmarkScored 2.6 on Harder read speechleaderboard
Aug 1, 2026ListedListed on LLMapfirst indexed by our pipeline
Aug 1, 2026BenchmarkScored 5406.7 on Open ASR RTFxleaderboard
Aug 1, 2026BenchmarkScored 6.7 on Open ASR WERleaderboard

Prices last checked 6h ago

What we do not know about this model yet

  • We hold no measured file for it, so all 3 sizes on this page are calculated from the parameter count.
  • Nothing we hold says whether an endpoint streams, so we do not show it either way.
04

Licence and identifiers

What the licence allowsCreative Commons Attribution 4.0, what it allows commercially, and the identifiers you need to pull this model — its Hugging Face repo, our slug and a machine-readable card.

Licence

Creative Commons Attribution 4.0

permissiveCommercial use allowed

Permissive content license: any use with attribution. Common for datasets and some model weights.

Identifiers

Modality record
audio->text
Catalogue slug
nvidia-parakeet-rnnt-0-6b

Machine-readable model card (omc.json) →

Something wrong on this page? Tell us