We value your privacy

We use cookies to enhance your browsing experience, serve personalized content and analyze our traffic. By clicking “Accept all” you consent to our use of cookies. Cookie Policy

More options
Skip to content
FacthoryFacthory
FacthoryFacthory
    • Workspace
    • Agent
    • Knowledge
    • Analytics
    • Office apps
    • Desktop
    All features
    What’s new
    Download
  • By work

    • Organizational Knowledge Risk & Gap Intelligence
    • Scientific Research & Experimentation
    • Autonomous Analytics & Insight Discovery
    • Workforce Readiness & Adaptive Learning
    • SOP Governance & Operational Standardization
    • Process Discovery & Continuous Optimization
    • All solutions

    By industry

    • Manufacturing
    • Automotive
    • Construction & Infrastructure
    • Real Estate
    • Public Sector
    • Life Sciences
    • All industries
  • Facthory for

    • Regulated industries
    • On-premises
    • Developers

    Discover

    • What’s new
    • Blog
    • Case studies
    • Webinars
    • Models

    Learn

    • Documentation
    • Help centre
  • Pricing
  • Download
Sign inGet started

A workspace for what your organization knows, with an agent that works on it.

Product

  • All features
  • Workspace
  • Agent
  • Knowledge
  • Analytics
  • Office apps
  • Desktop
  • Download
  • Pricing

Solutions

  • Organizational Knowledge Risk & Gap Intelligence
  • Scientific Research & Experimentation
  • Autonomous Analytics & Insight Discovery
  • Workforce Readiness & Adaptive Learning
  • SOP Governance & Operational Standardization
  • Process Discovery & Continuous Optimization
  • All solutions
  • All industries

Resources

  • Blog
  • Case studies
  • Webinars
  • Documentation
  • Help centre
  • Models
  • ROI calculator

Company

  • Who we are
  • Careers
  • Press hub
  • Brand kit
  • Philosophy
  • Contact us

Join our newsletter

Product news from Facthory, by email.

​

© 2026 Facthory. All rights reserved.
  • Privacy policy
  • Cookie policy
  • Terms and conditions
  • Imprint
  • Cookie settings
All modelsnvidia

parakeet-tdt-0.6b-v3

automatic-speech-recognition
627.1M parameters
cc-by-4.0
No file flagged

parakeet-tdt-0.6b-v3 is a 600-million-parameter multilingual automatic speech recognition (ASR) model designed for high-throughput speech-to-text transcription.

View on Hugging FaceCompare with other models

Fact sheet

Parameters

627.1M

Context length

not published

Architecture

ParakeetForTDT

Shape

not published

Weights stored as

float32

Task

automatic-speech-recognition

Languages

en, es, fr, de, bg, hr, cs, da, nl, et, fi, el, hu, it, lv, lt, mt, pl, pt, ro, sk, sl, sv, ru, uk

Built on

–

Released

4 Aug 2025

Last updated

5 Aug 2026

Downloads (last 30 days)

527,514

Licence and access

Licence

cc-by-4.0

Access

Weights download without approval

Security

Hugging Face security scan

No file flagged

Weight files by format

Safetensors: 1
Pickle-based: 0
GGUF: 1
Other formats: 0
Pickle-based files (.bin, .pt, .pth, .pkl, .ckpt) can run code when they are loaded; safetensors files cannot.

Facthory’s checks

Automated
Met
Safe weight formats

Every weight file is safetensors or GGUF.

Met
The hub’s malware and pickle scan

No file flagged.

Met
A declared licence

The card declares its licence.

Met
Published safety evaluations

The card reports safety or harm evaluations with their results.

Rated by Facthory’s crawler from the published facts; a person has not reviewed it yet.

GPU calculator

How much memory this model needs to serve at a quantization, context and number of concurrent sequences, which GPUs hold it, and how fast it can generate. Estimates from the model’s published size and shape, with the assumptions below.

Weights

BF16 / FP16 (16 bit)
​

Context per sequence: 8K tokens

Concurrent sequences: 1

KV cache

Memory

1.4 GB

Weights: 1.3 GB

KV cache: –

Runtime: 0.1 GB

The model’s shape is not published, so the KV cache is not included.

GPUMemoryGPUs neededPer sequenceAll sequences
NVIDIA L424 GB
1 ×
144 tokens/s144 tokens/s
NVIDIA L40S48 GB
1 ×
413 tokens/s413 tokens/s
NVIDIA RTX 6000 Ada48 GB
1 ×
459 tokens/s459 tokens/s
NVIDIA RTX PRO 6000 Blackwell96 GB
1 ×
857 tokens/s857 tokens/s
NVIDIA A100 80GB SXM80 GB
1 ×
976 tokens/s976 tokens/s
NVIDIA H100 SXM80 GB
1 ×
1,603 tokens/s1,603 tokens/s
NVIDIA H200 SXM141 GB
1 ×
2,296 tokens/s2,296 tokens/s
AMD Instinct MI300X192 GB
1 ×
2,536 tokens/s2,536 tokens/s
AMD Instinct MI325X256 GB
1 ×
2,871 tokens/s2,871 tokens/s
Assumptions
  • Runtime memory (activations, buffers) is 10% of weights and KV cache, plus 1 GB per GPU for the framework.
  • A GPU is filled to at most 90% of its memory; tensor parallelism uses 1, 2, 4 or 8 GPUs in one server.
  • Generation is limited by memory bandwidth at 60% of the vendor’s peak: each step reads the weights once and every sequence’s KV cache. A mixture-of-experts model reads only its active experts, so it runs faster than shown.
  • GPU memory and bandwidth are the vendors’ published figures.

Profile

Kind

Instruction model

Lineage

Original

Family

Parakeet TDT

Quantization

gguf

Uses

Speech
Translation
Conversation

In and out

Audio → Text

Written on 11 Nov 2026 by Workers AI (@cf/zai-org/glm-4.7-flash) from the model card and the hub’s metadata: a reading of the publisher’s text, not checked by Facthory.

Published benchmarks

As the model card publishes them, with the source the card names. Facthory has not run them; results are self-reported unless the hub verified them.

BenchmarkMetricResultSourceCounts toward
AMI (Meetings test)
ihm · test
Not counted: not on our list
Test WER11.31Model cardNot counted: not on our list
CoVoST2
de · test
Not counted: not on our list
Test WER (De)4.84Model cardNot counted: not on our list
CoVoST2
en · test
Not counted: not on our list
Test WER (En)6.8Model cardNot counted: not on our list
CoVoST2
es · test
Not counted: not on our list
Test WER (Es)3.41Model cardNot counted: not on our list
CoVoST2
et · test

EU AI Act

What the EU AI Act asks of a general-purpose AI model depends on who places it on the EU market and how. These facts bear on it:

  • Licence: cc-by-4.0
  • Access: Weights download without approval
  • Parameters: 627.1M
Automated
Rated by Facthory’s crawler from the published facts; a person has not reviewed it yet.

General-purpose AI model (Article 3(63))

Not indicated. Its task (automatic-speech-recognition) generates no text, audio, images or video, which the Commission’s indicative criterion asks for.

Articles 51 to 55, the obligations for general-purpose AI models, do not bind its provider.

Regulation (EU) 2024/1689, the EU AI Act · Source: General-Purpose AI Code of Practice, read 8 Nov 2026

Facthory’s EU AI Act score appears once a person has reviewed these checks. The articles we score against

Source: Read from Hugging Face on 19 Nov 2026, at commit 541d1f9. huggingface.co/nvidia/parakeet-tdt-0.6b-v3
How we rate
Not counted: not on our list
Test WER (Et)
22.04
Model card
Not counted: not on our list
CoVoST2
fr · test
Not counted: not on our list
Test WER (Fr)6.05Model cardNot counted: not on our list
CoVoST2
it · test
Not counted: not on our list
Test WER (It)3.69Model cardNot counted: not on our list
CoVoST2
lv · test
Not counted: not on our list
Test WER (Lv)38.36Model cardNot counted: not on our list
CoVoST2
nl · test
Not counted: not on our list
Test WER (Nl)6.5Model cardNot counted: not on our list
CoVoST2
pt · test
Not counted: not on our list
Test WER (Pt)3.96Model cardNot counted: not on our list
CoVoST2
ru · test
Not counted: not on our list
Test WER (Ru)3Model cardNot counted: not on our list
CoVoST2
sl · test
Not counted: not on our list
Test WER (Sl)31.8Model cardNot counted: not on our list
CoVoST2
sv · test
Not counted: not on our list
Test WER (Sv)20.16Model cardNot counted: not on our list
CoVoST2
uk · test
Not counted: not on our list
Test WER (Uk)5.1Model cardNot counted: not on our list
Earnings-22
test
Not counted: not on our list
Test WER11.42Model cardNot counted: not on our list
FLEURS
bg_bg · test
Not counted: not on our list
Test WER (Bg)12.64Model cardNot counted: not on our list
FLEURS
cs_cz · test
Not counted: not on our list
Test WER (Cs)11.01Model cardNot counted: not on our list
FLEURS
da_dk · test
Not counted: not on our list
Test WER (Da)18.41Model cardNot counted: not on our list
FLEURS
de_de · test
Not counted: not on our list
Test WER (De)5.04Model cardNot counted: not on our list
FLEURS
el_gr · test
Not counted: not on our list
Test WER (El)20.7Model cardNot counted: not on our list
FLEURS
en_us · test
Not counted: not on our list
Test WER (En)4.85Model cardNot counted: not on our list
FLEURS
es_419 · test
Not counted: not on our list
Test WER (Es)3.45Model cardNot counted: not on our list
FLEURS
et_ee · test
Not counted: not on our list
Test WER (Et)17.73Model cardNot counted: not on our list
FLEURS
fi_fi · test
Not counted: not on our list
Test WER (Fi)13.21Model cardNot counted: not on our list
FLEURS
fr_fr · test
Not counted: not on our list
Test WER (Fr)5.15Model cardNot counted: not on our list
FLEURS
hr_hr · test
Not counted: not on our list
Test WER (Hr)12.46Model cardNot counted: not on our list
FLEURS
hu_hu · test
Not counted: not on our list
Test WER (Hu)15.72Model cardNot counted: not on our list
FLEURS
it_it · test
Not counted: not on our list
Test WER (It)3Model cardNot counted: not on our list
FLEURS
lt_lt · test
Not counted: not on our list
Test WER (Lt)20.35Model cardNot counted: not on our list
FLEURS
lv_lv · test
Not counted: not on our list
Test WER (Lv)22.84Model cardNot counted: not on our list
FLEURS
mt_mt · test
Not counted: not on our list
Test WER (Mt)20.46Model cardNot counted: not on our list
FLEURS
nl_nl · test
Not counted: not on our list
Test WER (Nl)7.48Model cardNot counted: not on our list
FLEURS
pl_pl · test
Not counted: not on our list
Test WER (Pl)7.31Model cardNot counted: not on our list
FLEURS
pt_br · test
Not counted: not on our list
Test WER (Pt)4.76Model cardNot counted: not on our list
FLEURS
ro_ro · test
Not counted: not on our list
Test WER (Ro)12.44Model cardNot counted: not on our list
FLEURS
ru_ru · test
Not counted: not on our list
Test WER (Ru)5.51Model cardNot counted: not on our list
FLEURS
sk_sk · test
Not counted: not on our list
Test WER (Sk)8.82Model cardNot counted: not on our list
FLEURS
sl_si · test
Not counted: not on our list
Test WER (Sl)24.03Model cardNot counted: not on our list
FLEURS
sv_se · test
Not counted: not on our list
Test WER (Sv)15.08Model cardNot counted: not on our list
FLEURS
uk_ua · test
Not counted: not on our list
Test WER (Uk)6.79Model cardNot counted: not on our list
GigaSpeech
test
Not counted: not on our list
Test WER9.59Model cardNot counted: not on our list
LibriSpeech (clean)
other · test
Not counted: not on our list
Test WER1.93Model cardNot counted: not on our list
Multilingual LibriSpeech
dutch · test
Not counted: not on our list
Test WER (Nl)12.78Model cardNot counted: not on our list
Multilingual LibriSpeech
french · test
Not counted: not on our list
Test WER (Fr)4.97Model cardNot counted: not on our list
Multilingual LibriSpeech
italian · test
Not counted: not on our list
Test WER (It)10.08Model cardNot counted: not on our list
Multilingual LibriSpeech
polish · test
Not counted: not on our list
Test WER (Pl)7.28Model cardNot counted: not on our list
Multilingual LibriSpeech
portuguese · test
Not counted: not on our list
Test WER (Pt)7.5Model cardNot counted: not on our list
Multilingual LibriSpeech
spanish · test
Not counted: not on our list
Test WER (Es)4.39Model cardNot counted: not on our list
SPGI Speech
test · test
Not counted: not on our list
Test WER3.97Model cardNot counted: not on our list
Vox Populi
en · test
Not counted: not on our list
Test WER6.14Model cardNot counted: not on our list
tedlium-v3
release1 · test
Not counted: not on our list
Test WER2.75Model cardNot counted: not on our list

The capability score counts results the model’s authors publish, for the benchmarks our methodology lists, where at least two models in the catalog report the same benchmark, configuration and metric. Results a card attributes to the Open LLM Leaderboard are the leaderboard’s runs, and Epoch AI’s evaluations are Epoch’s: neither counts. How we score capability