Nemotron 3 Diarization needs about 0.6 GB of memory at Q8_0 (a 0.1 GB file) with an 8K context window, so it fits on a 4 GB GPU. This repository has 2 quants, from 0.1 GB (Q8_0) to 0.2 GB (BF16).
Nemotron 3 Diarization – speaker diarization up to eight speakers
Nemotron 3 Diarization is an open‑weight model that determines "who spoke when" in real‑world audio. It supports streaming and offline inference with a latency as low as 80 ms. The model handles up to eight speakers and returns speaker turns and timestamps.
Good at
- Streaming diarization
- Low‑latency speaker tracking
- Up to eight speakers
Limits
- No vision support
- Requires 16 kHz WAV input
- No transcription output
Best for
- Real‑time conference monitoring
- Podcast speaker separation
- Call‑center speaker analytics
Who it’s for. Developers building speaker‑aware audio applications or services that need to identify speaker turns without transcribing speech. The model is suitable for both commercial and non‑commercial use.
Quants & memory
| Quant | File | Est. memory | vs VRAM | Result | |
|---|---|---|---|---|---|
| Q8_0 | 0.1 GB | 0.6 GB | Fits | Get file → | |
| BF16 | 0.2 GB | 0.7 GB | Fits | Get file → |
For 8 GB VRAM use the Q8_0 file (0.11 GB, ~0.6 GB VRAM at 8k context). For 4 GB VRAM use the BF16 file (0.2 GB, ~0.7 GB VRAM at 8k context). Estimates are approximate.
What GPU do you need for Nemotron 3 Diarization?
These figures add the file size, an fp16 KV cache at an 8K context window and about 0.5 GB for the runtime. They are estimates, so leave a little headroom, and expect to use less context or a smaller quant if you are right at the limit.
- 4 GB VRAM: the best quant that fits is BF16 (0.2 GB file, about 0.7 GB in use).
Which Nemotron 3 Diarization quant should you choose?
Q8_0 (0.1 GB) is the middle option in this repository. Close to lossless. It costs roughly twice the memory of a 4-bit file, so it suits cases where you have plenty of VRAM.
New to quant names? Read GGUF quantization explained.
Specifications
Architecture from the GGUF header
Source
Frequently asked questions about Nemotron 3 Diarization
How much VRAM does Nemotron 3 Diarization need?
About 0.6 GB at Q8_0 with an 8K context window. That is the 0.1 GB file, plus the KV cache and roughly 0.5 GB of runtime overhead. Smaller quants need less and longer context needs more.
Can I run Nemotron 3 Diarization on an 8 GB GPU?
Yes, with BF16: the 0.2 GB file needs about 0.7 GB at 8K context. Larger quants in this repository do not fit in 8 GB.
What is the file size of Nemotron 3 Diarization GGUF?
It depends on the quant: Q8_0 is 0.1 GB, BF16 is 0.2 GB.
Where can I download Nemotron 3 Diarization GGUF?
The files are on Hugging Face in the audio-cpp/Nemotron-3-Diarization-GGUF repository. The download buttons on this page open the file on Hugging Face. ggufmodels.com does not host any model files.
What license does Nemotron 3 Diarization use?
Hugging Face lists the license as openmdw-1.1. Licenses can differ between the quantized files and the original model, so check the repository and the base model before commercial use.
Can I run Nemotron 3 Diarization on 4 GB VRAM?
Yes, using the BF16 quantization which uses about 0.7 GB VRAM at 8k context.
Does the model support more than eight speakers?
No, the model is limited to a maximum of eight speakers.
Is the model compatible with audio.cpp?
Yes, the GGUF weights are designed for native inference with audio.cpp.