Speech-to-text Sourced statement

Voxtral Mini 4B Realtime (2602) on NVIDIA GPU, 12–16 GB VRAM

Benchmark pending. We haven't measured this pair yet. Below is what the sources say about this hardware, quoted with links.

Mid-range. Most open TTS/STT models fit in fp16; LLM-based TTS at 3–4B may need 16 GB.

EvidenceSourced statement
Commercial useYes
Licenseapache-2.0
Stated memorySingle GPU with >= 16 GB memory (BF16, vLLM)
Device memory12–16 GB VRAM

Results

Benchmark pending. No measured numbers from us yet for Voxtral Mini 4B Realtime (2602) on 12–16 GB GPU. Get notified or send us your numbers (with the method).

What the sources say about 12–16 GB GPU

  • Model authors Model card: runs on a single GPU with >= 16 GB memory (BF16, vLLM). source
    Caveat: Needs the 16 GB end of this tier; a 12 GB card won't fit in BF16

Method

We haven't run this pair. Everything on this page is quoted from the linked sources, with the source type labelled (model authors vs third parties). Vendor numbers are often measured under favourable conditions.

Run it

We haven't verified a run command for this device yet. Start from the official instructions: https://huggingface.co/mistralai/Voxtral-Mini-4B-Realtime-2602.

Caveats

  • Needs the 16 GB end of this tier; a 12 GB card won't fit in BF16

New open speech models: will they run on your hardware?

One email a week: new open TTS/STT models, license gotchas, and what we measured on CPUs, Macs, Pis and consumer GPUs.