whatmodel.sh / stt / MAI-Transcribe-1.5
MAI-Transcribe-1.5
↗ share
+ contribute
MAI-Transcribe-1.5 is a closed-weights speech-to-text model from Microsoft AI, released June 2026.
Catalogue identity notes
Exact model_id 'mai-transcribe-1.5' recorded per run (koedesk, 2026-07-05). Predecessor of MAI-Transcribe-2; NOT the catalogue model mai-transcribe-2.
Model information sources
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 MAI-Transcribe-1.5, out of 176 with a score.
| # | model | score | where |
|---|---|---|---|
| 19 |  Granite 4.0 1B Speech | 62±3 | local |
| 20 | Granite Speech 5.0 470M TurboCTC | 62±4 | local |
| 21 | MAI-Transcribe-1.5 | 62±8 | api $0.0060/audio min |
| 22 |  Fish Audio Transcribe 1 | 62±5 | api $0.0060/audio min |
| 23 |  Inworld STT-1 | 62±6 | api $0.0025/audio min |
see all speech-to-text models · find the ones that fit your machine
$ what it's good at
other tests1 not in a task · 75% of its score
| test | result | vs the others in its run | counts |
|---|---|---|---|
| koedesk FLEURS first-200 · 9-language headline error mean | 3.6% ✎ | ▄·····▄··▄█▄ beats 5/5 | 75% |
$ where to run it
Through an API
api · Microsoft Azure · $0.0060/audio min · azure.microsoft.com ↗ · price sourceUSD per input audio minute: Azure Speech lists MAI Transcribe-1.5 at $0.36 per hour (÷60), US East standard pay-as-you-go (Speech to Text, Standard Transcription). Unlike MAI Transcribe-2, it carries no promotional footnote. Product: Azure AI Speech speech to text. API model: MAI-Transcribe-1.5. Settings: S0 pay-as-you-go, region=us-east. Price basis: published list rate (price embedded in the official page's regional price data), per-hour to per-minute scaling only. Exclusions: add-ons, commitment tiers, US Gov regions ($0.45/hr). Currency: USD. Microsoft Azure official pricing page, observed 2026-09-25.
$ how the score was worked out · koedesk STT Bench v1 (Track A… 75%
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 |
|---|---|---|
 koedesk STT Bench v1 (Track A, FLEURS first-200) | 1 | 75% |
| prior · pull toward the middle | 25% | |
| weighted average | = 62.2 |
$ 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