Gemma 4 E2B it
Google · released Mar 2, 2026 · google/gemma-4-E2B-it
Speech to textTranscribes a recording into words
- Type
- Open weightsApache License 2.0
- Languages
- 140
- Size
- 5.1B
Context measured in tokens
Our take
Written Sep 14, 2026Gemma 4 E2B it is Google's downloadable speech-to-text model with a permissive Apache licence and support for 140 languages. Its transcription speed is very fast, but its accuracy on every English condition we measure sits below the field median.
Pick this when you need broad language coverage from a permissive licence and can tolerate moderate English accuracy. Use it for offline batch work where speed matters more than word-for-word precision, or when licence terms outweigh raw accuracy. Skip it if you need reliable transcription of meetings, podcasts, accented speech, or clean read-aloud audio.
The case for it
- 140 languages supported — extremely broad coverage for a downloadable speech model.
- Very fast: processes an hour of audio in 19 seconds on benchmark hardware.
- Apache 2.0 licence allows commercial use, fine-tuning and redistribution.
The case against it
- Below-median accuracy on every English condition measured, with an overall word error rate more than double the field middle.
- Particularly weak on common real-world audio: podcasts, accented speech and meetings all score worse than most models measured.
- No live offers or pricing in our data, so you must self-host or source inference yourself.
How good is it?
An open speech-to-text model from Google for turning recordings into written text, though it trails most models on accuracy.
- turning spoken English into written textOpen ASR WER · 72nd of 76
- transcribing recordings of meetings in a roomRecorded meetings · 88th of 92
- transcribing speakers with a range of accentsAccented speech · 72nd of 76
- transcribing podcasts and video audioPodcasts and video · 87th of 92
- transcribing clear recordings of people reading aloudClean read speech · 90th of 92
TranscriptionTurning speech into text1 of 5Open ASR WER · 72nd of 76
88.4%
Misses roughly one word in 9, averaged over nine English test sets.
191×58th of 74
an hour of audio in 19 seconds, on the board's own hardware. Your machine will differ.
140
Stated by the leaderboard; we do not hold the list itself.
Percentage of words wrong on each set, lower better. Bars are scaled to this model's own worst case; the placing beneath each rate is against every model measured on that set.
The figures above come from the Open ASR Leaderboard, an independent public test that runs every model on the same recordings. It is the only measurement of transcription quality we know of, so there are no other scores to show.
Each of these is the same transcription job on a different kind of recording, so together they say where it holds up and where it slips — not how closely it follows an instruction.
Every published score for this model9 scoresEvery figure we hold, from 9 boards, with who ran it and a link to the source — including the boards no rating above is built on.
Can you run it yourself?
Comfortable fit
GeForce RTX 4090 · 24 GB
Room to spare. 18.2 GB spare means a 10% error in the size would not change the answer.
GeForce RTX 5090 · 32 GB
Room to spare. 26.2 GB spare means a 10% error in the size would not change the answer.
Apple M2 (8-core GPU, 8GB unified) · 8 GB
Room to spare. 1.4 GB spare means a 10% error in the size would not change the answer.
These cards answer whether Gemma 4 E2B it loads, not how fast it transcribes. Throughput figures for a transcription model come from its text decoder, so treat this as a fit answer rather than a speed one.
Memory use by level
Against a 24 GB card.
What is quantisation? →This model on every device we track71 devicesThe Q4 build most people download, on each device: what the weights come to, how much context the memory leaves, and whether it runs. Smallest device that runs it first. This is a fit answer: whether it loads, not how fast it transcribes.
When we formed this view
Recent changes
What moved
first indexed by our pipelineEach date is the day we first saw the change, or the day the maker announced it.
What we do not know about this model yet
- We hold no measured file for it, so all 3 sizes on this page are calculated from the parameter count.
- Nothing we hold says whether an endpoint streams, so we do not show it either way.
Licence and identifiers
What the licence allowsApache License 2.0, what it allows commercially, and the identifiers you need to pull this model — its Hugging Face repo, our slug and a machine-readable card.
Licence
Apache License 2.0
Fully permissive: commercial use, redistribution, and derivatives allowed. Requires attribution and a copy of the license. Includes an express patent grant.
Identifiers
- Hugging Face
- google/gemma-4-E2B-it
- Architecture
- Dense
- Takes in, gives back
- Audio in, text out
- Catalogue slug
- google-gemma-4-e2b-it