Parakeet TDT 0.6B v3
Parakeet TDT 0.6B v3 is a 600M-parameter speech-to-text model from NVIDIA, released under the CC BY 4.0 license. The reviewed source lists 25 European languages (25 explicitly enumerated in our record), and the model is designed for batch transcription, with live use only through a runtime or chunking wrapper. Among the common setups we test, it ranks highest for long recordings on a mid-range NVIDIA GPU (#2 of 26 in the matcher).
High-throughput multilingual ASR with 25 explicitly listed languages, word/segment timestamps and chunked streaming options. CC BY 4.0 attribution applies.
Key facts
| Provider | NVIDIA |
|---|---|
| Model family | Parakeet |
| Parameters | 600M |
| License | CC BY 4.0 · Attribution required — license text ↗ |
| Language coverage | 25 European languages · 25 enumerated in our record |
| Streaming | Streaming via a runtime or chunking wrapper |
| Timestamps | Yes — in the model or its runtime family |
| Speaker diarization | No — needs a separate component |
| Speech translation | No |
| Documented runtimes | NVIDIA NeMo, Hugging Face Transformers, NeMo-Speech.cpp — how to run it |
| Primary source | huggingface.co/nvidia/parakeet-tdt-0.6b-v3 ↗ |
| Last reviewed |
Hardware and workload fit
These are Speech Model Match's conservative 1–5 editorial ratings, relative to the other 19 reviewed models. They come from model size, official memory guidance and documented runtimes — not from a universal benchmark. How the ratings work.
| Dimension | Rating | Scale |
|---|---|---|
| CPU-only fit | 2/5 | |
| Apple Silicon fit | 3/5 | |
| NVIDIA GPU fit | 5/5 | |
| Speed | 5/5 | |
| Accuracy (relative) | 5/5 | |
| Low memory use | 3/5 | |
| Long recordings | 5/5 | |
| Live / low latency | 2/5 | |
| Subtitle work | 4/5 | |
| Edge / low-power devices | 2/5 |
Where it is strong
- NVIDIA GPU fit: rated 5/5
- Speed: rated 5/5
- Accuracy (relative): rated 5/5
- Long recordings: rated 5/5
- Subtitle work: rated 4/5
- Timestamps are available in the model or its runtime family
What to check first
- CPU-only fit: rated 2/5
- Live / low latency: rated 2/5
- Edge / low-power devices: rated 2/5
- No native speaker diarization — pair it with a separate diarization pipeline if you need speaker labels
- Commercial use is possible, but the license requires attribution.
How it ranks in common setups
Ranks come from the same deterministic matcher that runs on the homepage, applied to all 26 reviewed models. They are decision guidance, not accuracy claims.
| Setup | Rank | Fit score | Top match |
|---|---|---|---|
| Long recordings on an 8–12 GB NVIDIA GPU, balanced | #2 of 26 | 79/100 · Strong fit | Whisper turbo |
| Live captions or voice input (native streaming + timestamps, speed first) | #11 of 26 | 78/100 · Strong fit | Nemotron 3.5 ASR Streaming 0.6B |
| Subtitles with timing, accuracy first | #4 of 26 | 83/100 · Strong fit | Whisper medium |
| Meetings with speaker labels (diarization + timestamps) | #4 of 26 | 84/100 · Strong fit | VibeVoice-ASR 7B |
| CPU-only machine, low memory use | #12 of 26 | 61/100 · Possible fit | VibeVoice-ASR-BitNet |
| Apple Silicon with 8–16 GB unified memory, long recordings | #9 of 26 | 69/100 · Possible fit | Whisper small |
| Japanese live transcription (native streaming, speed first) | Not ranked (language not listed) | — | Nemotron 3.5 ASR Streaming 0.6B |
Supported languages
- Bulgarian
bg - Croatian
hr - Czech
cs - Danish
da - Dutch
nl - English
en - Estonian
et - Finnish
fi - French
fr - German
de - Greek
el - Hungarian
hu - Italian
it - Latvian
lv - Lithuanian
lt - Maltese
mt - Polish
pl - Portuguese
pt - Romanian
ro - Russian
ru - Slovak
sk - Slovenian
sl - Spanish
es - Swedish
sv - Ukrainian
uk
Compare coverage across every reviewed model in the language support matrix.
Community activity on Hugging Face
- Downloads, last 30 days
- 529,600
- Likes
- 1,270
- Repository last updated
- 5 Aug 2026
- Published on the Hub
- 4 Aug 2025
Figures from the Hugging Face Hub API, 11 Nov 2026. Source: huggingface.co/nvidia/parakeet-tdt-0.6b-v3 ↗. Downloads count file downloads, not active users, and do not affect the matcher's scores. Compare all models.
How to run Parakeet TDT 0.6B v3 locally
NVIDIA ships Parakeet for its NeMo toolkit; the model card also documents Transformers and a native C++ runtime. Runtimes below are the ones the official model card or repository documents for this model (checked November 2, 2026).
- NVIDIA NeMo ↗Framework · reference toolkit
- Hugging Face Transformers ↗Framework · installed from source, per the model card
- NeMo-Speech.cpp ↗Official package · NVIDIA's lightweight C++ runtime, q8_0 GGUF
Install command shown in the official huggingface.co page ↗:
pip install -U nemo_toolkit['asr']
Audio input: 16 kHz mono .wav or .flac. Up to 24 minutes per file with full attention, or up to 3 hours with local attention, according to the model card.
Official hardware guidance: At least 2 GB of RAM to load the model; more memory allows longer audio. Listed hardware: NVIDIA Ampere, Blackwell, Hopper and Volta GPUs, preferably on Linux. Source ↗
Whatever runtime you choose, measure speed and accuracy on your own audio with the benchmark tool; results differ a lot between runtimes, precision and hardware.
Before you deploy Parakeet TDT 0.6B v3
A printable checklist built from the facts on this page. Tick it off before putting the model into production.
- Language. Confirm your language is in the 25-language list on this page, and test any accents or code-switching you expect.
- License. CC BY 4.0: Give appropriate credit, provide a link to the license and indicate if changes were made. Do not add terms or technical measures that restrict others from doing what the license permits. License guide.
- Install. Start from the official command:
pip install -U nemo_toolkit['asr'](full run notes). - Audio input. 16 kHz mono .wav or .flac. Up to 24 minutes per file with full attention, or up to 3 hours with local attention, according to the model card.
- Hardware. At least 2 GB of RAM to load the model; more memory allows longer audio. Listed hardware: NVIDIA Ampere, Blackwell, Hopper and Volta GPUs, preferably on Linux.
- Features you rely on. For Parakeet TDT 0.6B v3, streaming only through a runtime or chunking wrapper; timestamps are documented; speaker labels need a separate diarization component. Test each one you need with real audio.
- Measure. Run 10–20 representative clips, score them with the WER calculator and record speed in the benchmark workspace before you commit.
Alternatives and related models
Same family: Parakeet Redux.
Similar fit profile: Distil-Whisper large-v3 (Hugging Face / Distil-Whisper, large-v3 distilled), Qwen3-ASR 0.6B (Qwen, 0.6B), Orukeet (Oruk AI, 627M). Similarity is computed from the ten fit ratings plus streaming and diarization support.
| Fact | Parakeet TDT 0.6B v3 | Parakeet Redux |
|---|---|---|
| Provider · size | NVIDIA · 600M | Moondream · 1.58-bit Parakeet 0.6B (178 MB) |
| License | CC BY 4.0 Attribution required | CC BY 4.0 Attribution required |
| Languages | 25 European languages | 25 European languages |
| Streaming | Via runtime | Batch-first |
| Timestamps | Yes | Not documented |
| Speaker diarization | Separate pipeline | Separate pipeline |
| CPU · Apple · NVIDIA fit | 2/5 · 3/5 · 5/5 | 5/5 · 4/5 · 3/5 |
| Live · long recordings | 2/5 · 5/5 | 2/5 · 4/5 |
| Main runtime | NVIDIA NeMo | Photon (moondream package) |
Head-to-head pages: Parakeet TDT 0.6B v3 vs Orukeet · Parakeet TDT 0.6B v3 vs Parakeet Redux
Compare Parakeet TDT 0.6B v3 side by side
Guides that cover Parakeet TDT 0.6B v3
- Parakeet v3 vs Nemotron 3.5 ASR Streaming
- Whisper vs Parakeet: which local ASR should you test first?
- Best local speech-to-text models in 2026
- Best local Whisper alternatives in 2026
- Best local STT models for an 8 GB GPU
- Best local speech-to-text models for multilingual use
Check Parakeet TDT 0.6B v3 for your setup
Pick your language, hardware and workload to see this model's fit score, the reasons behind it and where it ranks against all reviewed models. Everything runs in your browser.
Common questions
Can I use Parakeet TDT 0.6B v3 commercially?
Commercial use is possible, but the license requires attribution. The listed license is CC BY 4.0; read the current license text ↗ before shipping a product, because terms can change. The license guide compares the obligations of all 26 reviewed models.
Does Parakeet TDT 0.6B v3 support real-time streaming?
Not natively. Live use relies on a runtime or chunking wrapper around a batch model, which adds latency and can split words at chunk borders. If live latency is critical, compare it with a native streaming model such as Nemotron 3.5 ASR Streaming 0.6B.
Which languages does Parakeet TDT 0.6B v3 transcribe?
25 European languages. The full list of 25 explicitly enumerated languages is shown above.
Can Parakeet TDT 0.6B v3 tell speakers apart?
Not natively. Combine it with a dedicated diarization component and evaluate that component separately; the meetings guide covers this workflow.
Primary sources
Reviewed against primary sources on September 16, 2026. Model behavior, runtime support, language coverage and licenses can change; re-check the linked primary source before production use. See the full source directory.