GLM-5.2 is live. #1 throughput on OpenRouter, pay-per-token on FriendliAI. Try it today ➜
CohereLabs
Run this model inference on single tenant GPU with unmatched speed and reliability at scale.
Talk with our engineer to get a quote for reserved GPU instances with discounts.
Model provider
CohereLabs
Model tree
Base
this model
Modalities
Input
Audio
Output
Text
Pricing
Dedicated Endpoints
View detailsSupported Functionality
Model APIs
Dedicated Endpoints
Container
More information
Cohere Transcribe Arabic is supported natively in transformers. This is the recommended way to use the model for
offline inference. For online inference, see the vLLM integration example below.
bash
pip install transformers>=5.4.0 torch huggingface_hub soundfile librosa sentencepiece protobuf accelerate
Transcribe any audio file in a few lines:
python
from transformers import AutoProcessor, CohereAsrForConditionalGenerationfrom transformers.audio_utils import load_audiofrom huggingface_hub import hf_hub_downloadprocessor = AutoProcessor.from_pretrained("CohereLabs/cohere-transcribe-arabic-07-2026")model = CohereAsrForConditionalGeneration.from_pretrained("CohereLabs/cohere-transcribe-arabic-07-2026", device_map="auto")# Example: transcribe Arabic audioaudio_file = "your_audio.wav"audio = load_audio(audio_file, sampling_rate=16000)inputs = processor(audio, sampling_rate=16000, return_tensors="pt", language="ar")inputs.to(model.device, dtype=model.dtype)outputs = model.generate(**inputs, max_new_tokens=256)text = processor.decode(outputs, skip_special_tokens=True)print(text)
For audio longer than the feature extractor's max_audio_clip_s, the feature extractor automatically splits the waveform into chunks.
The processor reassembles the per-chunk transcriptions using the returned audio_chunk_index.
python
from transformers import AutoProcessor, CohereAsrForConditionalGenerationimport timeprocessor = AutoProcessor.from_pretrained("CohereLabs/cohere-transcribe-arabic-07-2026")model = CohereAsrForConditionalGeneration.from_pretrained("CohereLabs/cohere-transcribe-arabic-07-2026", device_map="auto")audio = load_audio("your_long_audio.wav", sampling_rate=16000)sr = 16000duration_s = len(audio) / srprint(f"Audio duration: {duration_s / 60:.1f} minutes")inputs = processor(audio=audio, sampling_rate=sr, return_tensors="pt", language="ar")audio_chunk_index = inputs.get("audio_chunk_index")inputs.to(model.device, dtype=model.dtype)start = time.time()outputs = model.generate(**inputs, max_new_tokens=256)text = processor.decode(outputs, skip_special_tokens=True, audio_chunk_index=audio_chunk_index, language="ar")[0]elapsed = time.time() - startrtfx = duration_s / elapsedprint(f"Transcribed in {elapsed:.1f}s — RTFx: {rtfx:.1f}")print(text)
The model also supports English. Specify language="en":
python
inputs = processor(audio, sampling_rate=16000, return_tensors="pt", language="en")inputs.to(model.device, dtype=model.dtype)outputs = model.generate(**inputs, max_new_tokens=256)text = processor.decode(outputs, skip_special_tokens=True)print(text)
For production serving we recommend running via vLLM following the instructions below.
First install vLLM (refer to vLLM installation instructions):
bash
uv venv --python 3.12 --seedsource .venv/bin/activateuv pip install -U vllm==0.19.0 --torch-backend=autouv pip install vllm[audio]uv pip install librosa
Start vLLM server
bash
vllm serve CohereLabs/cohere-transcribe-arabic-07-2026 --trust-remote-code
Send request
bash
curl -v -X POST http://localhost:8000/v1/audio/transcriptions \-H "Authorization: Bearer $VLLM_API_KEY" \-F "file=@$(realpath ${AUDIO_PATH})" \-F "model=CohereLabs/cohere-transcribe-arabic-07-2026"
Link to the live leaderboard: Open Universal Arabic ASR Leaderboard.
For more details and results:
Cohere Transcribe Arabic demonstrates strong transcription accuracy for Arabic and English. As a dedicated speech recognition model, it benefits from efficient inference via the Conformer encoder-decoder architecture.
Single language. The model performs best when remaining in-distribution of a single, pre-specified language. It does not feature explicit, automatic language detection and exhibits inconsistent performance on code-switched audio.
Timestamps/Speaker diarization. The model does not feature either of these.
Silence. Like most AED speech models, Cohere Transcibe Arabic is eager to transcribe, even non-speech sounds. The model benefits from prepending a noise gate or VAD (voice activity detection) model in order to prevent low-volume, floor noise from turning into hallucinations.
For errors or additional questions about details in this model card, contact labs@cohere.com or raise an issue.
Terms of Use: We hope that the release of this model will make community-based research efforts into Arabic speech more accessible. This model is governed by an Apache 2.0 license.