Cohere Transcribe 03-2026
Cohere Transcribe 03-2026 is a 2B-parameter speech-to-text model from Cohere Labs, released under the Apache-2.0 license. The reviewed source lists 14 languages (14 explicitly enumerated in our record), and the model is designed for batch-first transcription. It is a specialist rather than a general pick: its best placement in the common setups we test is #12 of 26, for long recordings on a mid-range NVIDIA GPU. Use the matcher with your own language and hardware to see where it fits.
2B open-weights ASR model with strong published accuracy across 14 listed languages. The Hugging Face repository is gated even though the model is Apache-2.0.
Key facts
| Provider | Cohere Labs |
|---|---|
| Model family | Cohere Transcribe |
| Parameters | 2B |
| License | Apache-2.0 · Permissive — license text ↗ |
| Language coverage | 14 listed languages · 14 enumerated in our record |
| Streaming | Batch-first (no streaming path in the reviewed source) |
| Timestamps | Limited or runtime-dependent |
| Speaker diarization | No — needs a separate component |
| Speech translation | No |
| Documented runtimes | Hugging Face Transformers ≥ 5.4, vLLM, mlx-audio — how to run it |
| Primary source | huggingface.co/CohereLabs/cohere-transcribe-03-2026 ↗ |
| 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 | 1/5 | |
| Apple Silicon fit | 2/5 | |
| NVIDIA GPU fit | 5/5 | |
| Speed | 3/5 | |
| Accuracy (relative) | 5/5 | |
| Low memory use | 1/5 | |
| Long recordings | 5/5 | |
| Live / low latency | 2/5 | |
| Subtitle work | 4/5 | |
| Edge / low-power devices | 1/5 |
Where it is strong
- NVIDIA GPU fit: rated 5/5
- Accuracy (relative): rated 5/5
- Long recordings: rated 5/5
- Subtitle work: rated 4/5
What to check first
- No streaming path in the reviewed source — plan for file/batch transcription
- CPU-only fit: rated 1/5
- Apple Silicon fit: rated 2/5
- Low memory use: rated 1/5
- Live / low latency: rated 2/5
- Edge / low-power devices: rated 1/5
- No native speaker diarization — pair it with a separate diarization pipeline if you need speaker labels
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 | #12 of 26 | 71/100 · Possible fit | Whisper turbo |
| Live captions or voice input (native streaming + timestamps, speed first) | #24 of 26 | 50/100 · Weak fit | Nemotron 3.5 ASR Streaming 0.6B |
| Subtitles with timing, accuracy first | #12 of 26 | 73/100 · Strong fit | Whisper medium |
| Meetings with speaker labels (diarization + timestamps) | #12 of 26 | 74/100 · Strong fit | VibeVoice-ASR 7B |
| CPU-only machine, low memory use | #21 of 26 | 51/100 · Weak fit | VibeVoice-ASR-BitNet |
| Apple Silicon with 8–16 GB unified memory, long recordings | #19 of 26 | 61/100 · Possible fit | Whisper small |
| Japanese live transcription (native streaming, speed first) | #16 of 26 | 50/100 · Weak fit | Nemotron 3.5 ASR Streaming 0.6B |
Supported languages
- Arabic
ar - Dutch
nl - English
en - French
fr - German
de - Greek
el - Italian
it - Japanese
ja - Korean
ko - Mandarin Chinese
zh - Polish
pl - Portuguese
pt - Spanish
es - Vietnamese
vi
Compare coverage across every reviewed model in the language support matrix.
Community activity on Hugging Face
- Downloads, last 30 days
- 106,517
- Likes
- 1,238
- Repository last updated
- 7 Oct 2026
- Published on the Hub
- 24 Mar 2026
Figures from the Hugging Face Hub API, 11 Nov 2026. Source: huggingface.co/CohereLabs/cohere-transcribe-03-2026 ↗. Downloads count file downloads, not active users, and do not affect the matcher's scores. Compare all models.
How to run Cohere Transcribe 03-2026 locally
Cohere documents native Transformers support and recommends vLLM for production serving. Runtimes below are the ones the official model card or repository documents for this model (checked November 2, 2026).
- Hugging Face Transformers ≥ 5.4 ↗Framework · native support
- vLLM ↗Serving engine · recommended for production
- mlx-audio ↗Third-party runtime · Apple Silicon, listed on the model card
Install command shown in the official huggingface.co page ↗:
pip install transformers>=5.4.0 torch huggingface_hub soundfile librosa sentencepiece protobuf accelerate
Audio input: Audio is resampled to 16 kHz and stereo is averaged to mono automatically. You must pass the language: the model has no automatic language detection.
The weights are gated: Hugging Face asks you to share contact information before downloading.
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 Cohere Transcribe 03-2026
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 14-language list on this page, and test any accents or code-switching you expect.
- License. Apache-2.0: Give recipients a copy of the license, keep copyright and attribution notices, carry forward any NOTICE file and mark files you changed. The patent license ends if you start patent litigation over the work. License guide.
- Install. Start from the official command:
pip install transformers>=5.4.0 torch huggingface_hub soundfile librosa sentencepiece protobuf accelerate(full run notes). - Audio input. Audio is resampled to 16 kHz and stereo is averaged to mono automatically. You must pass the language: the model has no automatic language detection.
- Hardware. No vendor memory figure is published; our ratings are CPU 1/5, Apple 2/5, NVIDIA 5/5.
- Features you rely on. For Cohere Transcribe 03-2026, no streaming path is documented; timestamps are limited or runtime-dependent; 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
Similar fit profile: Qwen3-ASR 1.7B (Qwen, 1.7B), VibeVoice-ASR 7B (Microsoft, 9B total), Whisper medium (OpenAI, 769M). Similarity is computed from the ten fit ratings plus streaming and diarization support.
| Fact | Cohere Transcribe 03-2026 | Qwen3-ASR 1.7B | VibeVoice-ASR 7B |
|---|---|---|---|
| Provider · size | Cohere Labs · 2B | Qwen · 1.7B | Microsoft · 9B total |
| License | Apache-2.0 Permissive | Apache-2.0 Permissive | MIT Permissive |
| Languages | 14 listed languages | 30 languages + 22 Chinese dialects | 51 listed languages |
| Streaming | Batch-first | Via runtime | Batch-first |
| Timestamps | Limited | Separate aligner | Yes |
| Speaker diarization | Separate pipeline | Separate pipeline | Native |
| CPU · Apple · NVIDIA fit | 1/5 · 2/5 · 5/5 | 1/5 · 3/5 · 5/5 | 1/5 · 2/5 · 4/5 |
| Live · long recordings | 2/5 · 5/5 | 3/5 · 5/5 | 1/5 · 5/5 |
| Main runtime | Hugging Face Transformers ≥ 5.4 | qwen-asr package | microsoft/VibeVoice |
Head-to-head pages: Whisper turbo vs Cohere Transcribe 03-2026
Compare Cohere Transcribe 03-2026 side by side
Check Cohere Transcribe 03-2026 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 Cohere Transcribe 03-2026 commercially?
Permissive license; commercial use is allowed under the listed terms. The listed license is Apache-2.0; 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 Cohere Transcribe 03-2026 support real-time streaming?
No streaming path is documented in the reviewed source. Treat it as a file/batch transcription model.
Which languages does Cohere Transcribe 03-2026 transcribe?
14 listed languages. The full list of 14 explicitly enumerated languages is shown above.
Can Cohere Transcribe 03-2026 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.