# Cohere Transcribe Review

> Cohere Transcribe review: Enterprise-grade open ASR for accurate batch transcription. Verified pricing, features, and the strongest alternatives in.

![Cohere Transcribe logo](https://www.versusref.com/logos/cohere-transcribe.png)

Enterprise-grade open ASR for accurate batch transcription

Among the 43 speech-to-text tools we track, Cohere Transcribe has the 35th-widest language coverage.

See pricing · facts verified Jul 20, 2026

## What we know about Cohere Transcribe

Cohere Transcribe is a speech-to-text apis platform: enterprise-grade open ASR for accurate batch transcription. This profile tracks every Cohere Transcribe fact we have verified, each linked to a primary source and dated.

Cohere Transcribe does not publish a public per-minute rate we have been able to verify, so the price figure here stays blank until we can confirm one. We would rather show nothing than a guessed number; if you have a current Cohere Transcribe quote, it is the fastest way for us to close that gap.

On capabilities, Cohere Transcribe covers soc 2 type ii and self-host / on-prem option, and does not offer language auto-detection, word-level timestamps, and custom vocabulary / keyterm boosting. Each of those is verified against Cohere Transcribe's own docs or dashboard, not marketing copy.

For compliance, with Cohere Transcribe: SOC 2 Type II is in place. If you are in a regulated space, confirm the current posture with Cohere Transcribe before you commit, since these change plan by plan.

Placed against the 43 speech-to-text tools we track, Cohere Transcribe's strongest showing is the 35th-widest language coverage - a spread worth weighing against your own priorities.

In total we track 27 verified facts for Cohere Transcribe today, each linking the primary source it came from so you can check our work - and vendor claims we have not measured ourselves are labeled as such on the Cohere Transcribe fact sheet below.

## Fact sheet

### Pricing

| Fact | Value | Verified | Source |
| --- | --- | --- | --- |
| Pricing model | hybrid | Jul 20 | [source](https://cohere.com/blog/transcribe) |
| Free tier quota | Free rate-limited API; no minute quota published | Jul 20 | [source](https://cohere.com/blog/transcribe) |
| Published volume discounts | Longer-term Model Vault commitments discounted | Jul 20 | [source](https://cohere.com/blog/transcribe) |

### Capabilities

| Fact | Value | Verified | Source |
| --- | --- | --- | --- |
| WER (third-party benchmark) | 4.57 % WER | Jul 20 | [source](https://artificialanalysis.ai/speech-to-text) |
| WER (vendor-claimed) | ~5.42 % WER | Jul 20 | [source](https://huggingface.co/CohereLabs/cohere-transcribe-03-2026) |
| Languages supported | 14 languages | Jul 20 | [source](https://docs.cohere.com/docs/transcribe) |
| Language auto-detection | ✗  No | Jul 20 | [source](https://docs.cohere.com/docs/transcribe) |
| Speaker diarization | ✗  Not available | Jul 20 | [source](https://docs.cohere.com/docs/transcribe) |
| PII redaction | ✗  Not available | Jul 20 | [source](https://docs.cohere.com/docs/transcribe) |
| Word-level timestamps | ✗  No | Jul 20 | [source](https://docs.cohere.com/docs/transcribe) |
| Custom vocabulary / keyterm boosting | ✗  No | Jul 20 | [source](https://docs.cohere.com/docs/transcribe) |
| Entity detection | ✗  No | Jul 20 | [source](https://docs.cohere.com/docs/transcribe) |
| Sentiment analysis | ✗  No | Jul 20 | [source](https://docs.cohere.com/docs/transcribe) |
| Summarization endpoint | ✗  No | Jul 20 | [source](https://docs.cohere.com/docs/transcribe) |
| Speech translation | ✗  No | Jul 20 | [source](https://docs.cohere.com/docs/transcribe) |
| Priced audio-intelligence add-ons | None - transcription-only model | Jul 20 | [source](https://docs.cohere.com/docs/transcribe) |

### Compliance & trust

| Fact | Value | Verified | Source |
| --- | --- | --- | --- |
| SOC 2 Type II | ✓  Yes | Jul 20 | [source](https://cohere.com/security) |
| Self-host / on-prem option | ✓  Yes | Jul 20 | [source](https://cohere.com/transcribe) |
| Model weights license | Apache 2.0 | Jul 20 | [source](https://huggingface.co/CohereLabs/cohere-transcribe-03-2026) |

### Build experience

| Fact | Value | Verified | Source |
| --- | --- | --- | --- |
| Official SDKs | Python (cohere SDK) | Jul 20 | [source](https://docs.cohere.com/v2/docs/audio-transcription-quickstart) |
| Max file size / duration | 25MB max file; duration not specified | Jul 20 | [source](https://docs.cohere.com/v2/docs/audio-transcription-quickstart) |
| Supported audio formats | FLAC, MP3, MPEG, MPGA, OGG, WAV | Jul 20 | [source](https://docs.cohere.com/v2/docs/audio-transcription-quickstart) |
| Websocket streaming API | ✗  No | Jul 20 | [source](https://docs.cohere.com/docs/transcribe) |

### Commercial

| Fact | Value | Verified | Source |
| --- | --- | --- | --- |
| Model size (parameters) | 2B | Jul 20 | [source](https://huggingface.co/CohereLabs/cohere-transcribe-03-2026) |
| Hardware to self-host | Small GPU; exact VRAM not published | Jul 20 | [source](https://cohere.com/transcribe) |
| Hosted API available | Yes - free rate-limited API + Model Vault | Jul 20 | [source](https://cohere.com/blog/transcribe) |
| Project maintenance status | Active - released 2026-03; Arabic 2026-07 | Jul 20 | [source](https://docs.cohere.com/v2/docs/audio-transcription-quickstart) |

## Cohere Transcribe head-to-head

| Comparison | Record | Per use case |
| --- | --- | --- |
| [Cohere Transcribe vs Deepgram](https://www.versusref.com/stt/cohere-transcribe-vs-deepgram/) | won 1 · lost 5 · tied 1 | Call Centers: lost; Developers: lost; Dictation: tie; Medical: lost; Meetings: lost; Self-Hosted: won; Voice Agents: lost |

Source: https://www.versusref.com/stt/tools/cohere-transcribe/