ai-models rust webgpu speech-recognition

Rust implementation of Mistral's Voxtral Mini 4B Realtime runs in your browser

GitHub - TrevorS/voxtral-mini-realtime-rs

github.com

February 10, 2026

4 min read

🔥🔥🔥🔥🔥

65/100

Summary

Voxtral Mini Realtime is a streaming speech recognition model implemented in pure Rust, utilizing the Burn ML framework. It operates natively in the browser via WASM and WebGPU, with a Q4 GGUF quantized version available for client-side execution.

Key Takeaways

The Voxtral Mini 4B Realtime model is implemented in pure Rust and runs natively in the browser using WASM and WebGPU.
The model can transcribe audio files and supports a Q4 GGUF quantized path that is approximately 2.5 GB in size.
A hosted demo is available on HuggingFace Spaces, allowing users to try the model without local setup.
The implementation addresses multiple constraints, including a 2 GB allocation limit and a 4 GB address space for running the model in a browser tab.

Read original article

View story →

Coverage & discussions

Hacker News · github.com

Rust implementation of Mistral's Voxtral Mini 4B Realtime runs in your browser ↗

Community discussions

Hacker News discussion ↗

Historical count: 5 · 5 saved in sample · Historical sample

Analysis ready

Community Sentiment

Mixed

Positives

The Rust implementation of Voxtral Mini 4B demonstrates impressive capabilities by running directly in the browser, showcasing the potential for real-time AI applications.
User experiences indicate that the model can effectively transcribe speech, with improvements noted in subsequent tests, highlighting its evolving accuracy.

Concerns

Several users encountered runtime errors and performance issues, suggesting that the implementation may not be stable across different environments.
One user reported poor transcription quality, which raises concerns about the model's reliability and effectiveness in diverse scenarios.