# Voxtral Mini 4B Realtime

17 Hacker News commenters recommended Voxtral Mini 4B Realtime and 7 criticized it, November 2025 to November 2026. It was recommended in 3 of 13 months.

**Verdict:** Open-weights speech-to-text model under Apache 2.0 that people run locally for real-time dictation. The most common caveat is its size: a 2.5GB download and memory footprint, and it is too slow on some machines.

## Why people recommend it

- Real-time transcription on capable hardware (3 comments): "the model is definitely real time (able to transcribe audio faster than the length of the audio)" ([source](https://news.ycombinator.com/item?id=46983602)) "it is realtime on my DGX Spark with an RTF of .416" ([source](https://news.ycombinator.com/item?id=46983678)) "I'm using mistral voxtral realtime daily – it's great." ([source](https://news.ycombinator.com/item?id=47950105))
- Open weights, transparent, on-premise use (3 comments): "From a cybersecurity perspective, this project is impressive not just for performance, but for transparency." ([source](https://news.ycombinator.com/item?id=46958837)) "open-models running on-premise. Preferred by users and businesses." ([source](https://news.ycombinator.com/item?id=46956061)) "The Apache 2.0 license on Realtime is the buried lede." ([source](https://news.ycombinator.com/item?id=46897883))
- Used for daily dictation (3 comments): "I built a macOS dictation app on top of Voxtral Realtime" ([source](https://news.ycombinator.com/item?id=47149479)) "I'm using mistral voxtral realtime daily – it's great." ([source](https://news.ycombinator.com/item?id=47950105)) "Not perfect dictation, but close enough" ([source](https://news.ycombinator.com/item?id=46955491))
- Causal encoder, accurate for dictation (2 comments): "The thing that sold me on Voxtral Realtime over Whisper-based models for dictation is the causal encoder." ([source](https://news.ycombinator.com/item?id=47178784)) "At 480ms delay it matches offline models on accuracy (FLEURS benchmark)" ([source](https://news.ycombinator.com/item?id=47138004))

## What people warn about

- Too slow for real-time use (5 comments): "Given that it took 19.64 mins to transcribe the 11 second sample wav" ([source](https://news.ycombinator.com/item?id=46956410)) "This isn't even close to realtime on M4 Max." ([source](https://news.ycombinator.com/item?id=46956227)) "way too slow for anything close to realtime" ([source](https://news.ycombinator.com/item?id=46974548))
- Large 2.5GB download and memory (3 comments): "I don't want to live in a world where its normal for a website to download 2.5Gb in the background to run something" ([source](https://news.ycombinator.com/item?id=46957585)) "this is not yet practical for browser use due to it's (at least) 2.5GB memory footprint" ([source](https://news.ycombinator.com/item?id=46958728)) "it is too big for local inference" ([source](https://news.ycombinator.com/item?id=46958785))
- Fails to load or run on some machines (3 comments): "init failed: Worker error: Uncaught RuntimeError: unreachable" ([source](https://news.ycombinator.com/item?id=46954831)) "seems broken on my machine (Firefox, Asahi Linux, M1 Pro)" ([source](https://news.ycombinator.com/item?id=46954949)) "okay... so I cannot get this to run on my mac" ([source](https://news.ycombinator.com/item?id=46957722))

## Alternatives hackers recommend

- [OpenAI Whisper](https://hackerapproved.com/p/whisper.md): 164 recommend
- [NVIDIA Parakeet](https://hackerapproved.com/p/parakeet.md): 144 recommend
- [Mistral Voxtral](https://hackerapproved.com/p/voxtral-transcribe-2.md): 69 recommend
- [Moonshine open-weights STT models](https://hackerapproved.com/p/moonshine.md): 37 recommend
- [whisper.cpp](https://hackerapproved.com/p/whisper-cpp.md): 33 recommend

## Cite as

Hacker Approved, "Voxtral Mini 4B Realtime", https://hackerapproved.com/p/voxtral-mini-4b-realtime/ (counts from 1,795,640 Hacker News comments, November 2025 to November 2026).