whatmodel.sh / stt / Ink-Whisper
Ink-Whisper
↗ share
+ contribute
Ink-Whisper is a closed-weights speech-to-text model from Cartesia, released June 2025. Licence: proprietary.
Catalogue identity notes
Cartesia's earlier streaming STT model, API id ink-whisper (dated snapshot ink-whisper-2025-06-04, release date June 4, 2025, status Stable, per Cartesia's Older Models page). Cartesia describes it as delivering 'higher accuracy and lower latency than baseline Whisper'; architecture and size not stated. No per-model price is published separately from Ink-2's credit rate, so no price is recorded.
Score within the Open ASR English short-form WER (March 2026) comparison set only.
$ how it compares
The speech-to-text models ranked either side of Ink-Whisper, out of 176 with a score, then more from Cartesia.
| # | model | score | where | ||
|---|---|---|---|---|---|
| 86 |  SeamlessM4T v2 Large | 55±4 | local | ||
| 87 |  Qwen3-ASR-0.6B | 55±1 | local | ||
| 88 | Ink-Whisper | 55±3 | api price unknown | ||
| 89 |  Breeze ASR 25 | 54±4 | local | ||
| 90 |  Parakeet CTC 0.6B | 54±3 | local | ||
| more from Cartesia | |||||
| 7 | Ink-2 | 68±3 | api $0.0065/audio min | ||
see all speech-to-text models · find the ones that fit your machine
$ what it's good at
other tests3 not in a task · 97% of its score
| test | result | vs the others in its run | counts |
|---|---|---|---|
| Pipecat STT Benchmark · pooled semantic WER (1000 smart-turn samples, Sonnet 5.5 judge) | 3.8% ✎ | ▃·▃▅·▅▅·▃▅██ beats 1/16 | 72% |
| verli STT benchmark · LibriSpeech test-clean WER | 2.0% ✎ | ▃·····▃▅█▃▅▅ beats 4/11 | 17% |
| verli STT benchmark · Earnings-22 WER (2,741 utt.) | 11.6% ✎ | ▃·▃▅··▃▃▅▅█▃ beats 7/13 | 7.3% |
$ where to run it
Through an API
api · price unknown · docs.cartesia.ai ↗
$ how the score was worked out · Pipecat STT Benchmark · seman… 72%, verliapp/stt-benchmark · inde… 24%
Each result implies a score on its own, after adjusting for how hard its test is and who ran it. The score is their weighted average, with a small pull toward the middle for thinly tested models. Each test's weight is its counts column above; here it's added up by who ran the tests. How it works
| source | results | share of the score |
|---|---|---|
 Pipecat STT Benchmark · semantic WER results table (pipecat-ai/stt-benchmark README) | 1 | 72% |
 verliapp/stt-benchmark · independent STT accuracy benchmark | 2 | 24% |
| prior · pull toward the middle | 3% | |
| weighted average | = 54.6 |
$ improve this page
Use a short GitHub form or edit the data through a pull request. Every number needs a source; here's which ones we can use.
data updated · sources and licences · json