BroadNet Transcribe API
Updated
BroadNet Transcribe is a speech-to-text model based on Cohere Transcribe, served by BroadNet Inference at https://inference.broadnet.ai/v1/audio/transcriptions as broadnet/cohere-transcribe. It costs $0.002 per audio minute, accepts WAV or MP3, and no audio or transcript is retained. It can use the starting credit.
At a glance
- Audio (speech to text)
- Arabic
- 14 languages
- WAV or MP3 up to 25 MiB per file
- OpenAI-compatible /v1/audio/transcriptions
- BroadNet measures 5.5 WER on FLEURS English and 3.2 on FLEURS Spanish

| Model ID | broadnet/cohere-transcribe |
|---|---|
| Endpoint | /v1/audio/transcriptions |
| Supported inputs | Audio: WAV or MP3, up to 26,214,400 bytes (25 MiB) per file |
| Price | $0.002 per audio minute |
| Languages | English, Spanish, French, German, Italian, Portuguese, Greek, Dutch, Polish, Chinese (Mandarin), Japanese, Korean, Vietnamese, Arabic (language: en, es, fr, de, it, pt, el, nl, pl, zh, ja, ko, vi, ar) |
| Pay-as-you-go limits | 30 requests a minute, 4 in flight per key |
| Access | Starting credit eligible |
Prices on this page are from BroadNet's price feed snapshot dated , not checked live here. See official pricing for current rates.
official pricingReport a wrong price or number (email to info@broadnet.ai, with this page's link filled in).
What does BroadNet publish about BroadNet Transcribe?
From the BroadNet Transcribe page on broadnet.ai, checked 15 November 2026. Measurements are BroadNet's, on its hardware, unless a public leaderboard is named; we have not reproduced them.
| Model | Cohere Transcribe 03-2026, 2B parameters |
|---|---|
| Languages | 14, including English, Spanish, French, German, Chinese, Japanese, Korean and Vietnamese |
| Input | WAV and MP3 audio |
| Output | Text in the spoken language; JSON or plain text |
| Hardware | NVIDIA GPUs operated by BroadNet in the United States |
| Licence | Apache-2.0 (Cohere Labs) |
BroadNet's measurements
- FLEURS English 5.5 and FLEURS Spanish 3.2 word error rate.
- 113 times realtime on one stream, 217 times batched, on English audio.
- For Arabic audio, BroadNet recommends its Arabic model instead.
Base weights: CohereLabs/cohere-transcribe-03-2026 on Hugging Face.
How BroadNet chooses and serves models: its GPU hardware and serving approach and the models it evaluated and ruled out.
Published benchmark results
BroadNet did not run these benchmarks; results are as published by the Cohere model card and may use different settings. BroadRouter is BroadNet's site. They refer to the published model CohereLabs/cohere-transcribe-03-2026; not measured on BroadNet's serving.
| Benchmark | Result | Setting or note | Source |
|---|---|---|---|
| English ASR Leaderboard: average WER | 5.42 | Word error rate, lower is better; leaderboard as of 26 March 2026, as quoted on the card | Cohere model card, checked 2026-11-16 |
| English ASR Leaderboard: LibriSpeech clean WER | 1.25 | Word error rate, lower is better; leaderboard as of 26 March 2026, as quoted on the card | Cohere model card, checked 2026-11-16 |
| English ASR Leaderboard: LibriSpeech other WER | 2.37 | Word error rate, lower is better; leaderboard as of 26 March 2026, as quoted on the card | Cohere model card, checked 2026-11-16 |
| English ASR Leaderboard: Earnings 22 WER | 10.84 | Word error rate, lower is better; leaderboard as of 26 March 2026, as quoted on the card | Cohere model card, checked 2026-11-16 |
| English ASR Leaderboard: AMI WER | 8.15 | Word error rate, lower is better; leaderboard as of 26 March 2026, as quoted on the card | Cohere model card, checked 2026-11-16 |
Side by side: BroadNet Transcribe vs BroadNet Transcribe Arabic, with prices, languages and monthly costs.
Call it in Python
Install the OpenAI SDK with pip install openai, export BROADNET_API_KEY, then run:
from openai import OpenAI
import os
client = OpenAI(base_url="https://inference.broadnet.ai/v1", api_key=os.environ["BROADNET_API_KEY"])
with open("voice-note.wav", "rb") as audio:
result = client.audio.transcriptions.create(
model="broadnet/cohere-transcribe",
file=audio,
language="en", # always send the language
)
print(result.text)Or with curl
curl https://inference.broadnet.ai/v1/audio/transcriptions \
-H "Authorization: Bearer $BROADNET_API_KEY" \
-F model=broadnet/cohere-transcribe \
-F language=en \
-F file=@voice-note.wavGuides for BroadNet Transcribe
Use BroadNet Transcribe in the console
The BroadNet console on inference.broadnet.ai manages accounts and keys. It does not receive your choice from this page, so:
- Sign in or create an account.
- Create a pay-as-you-go key; the starting credit covers this model.
- Send this model ID as
modelto/v1/audio/transcriptions:broadnet/cohere-transcribe
Open the BroadNet consoleEstimate the costRead the API reference
Compare BroadNet Transcribe
Questions
What is the model ID for BroadNet Transcribe?
Pass broadnet/cohere-transcribe as model to /v1/audio/transcriptions. The base URL is https://inference.broadnet.ai/v1.
How much does BroadNet Transcribe cost?
$0.002 per audio minute, from your account balance.
Which languages does BroadNet Transcribe support?
English, Spanish, French, German, Italian, Portuguese, Greek, Dutch, Polish, Chinese (Mandarin), Japanese, Korean, Vietnamese, Arabic. Always send the language field; without it Arabic can come back transliterated into Latin letters.
Can I use BroadNet Transcribe without a top-up?
Yes. It can use the starting credit for new accounts.
Where does BroadNet Transcribe run?
BroadNet's model page says: NVIDIA GPUs operated by BroadNet in the United States. Its privacy statement adds that processing locations for data-residency needs are agreed with BroadNet before you send data.
How does BroadNet Transcribe score on benchmarks?
As published, the Cohere model card reports 5.42 on English ASR Leaderboard: average WER, 1.25 on English ASR Leaderboard: LibriSpeech clean WER and 2.37 on English ASR Leaderboard: LibriSpeech other WER (checked 2026-11-16). WER is word error rate: lower is better. BroadNet did not run these benchmarks, and sources may use different settings, so the figures are not directly comparable. Every published result we found, with its setting and source.
Is my data stored?
Inference content is not retained: audio and transcripts are processed for the request only. Necessary account, billing and security metadata, such as timings, token counts and cost, is retained as described in BroadNet's privacy statement. Read how.