NVIDIA · Nemotron
Nemotron 3 Diarization
Verified by Badgr · RunNemotron 3 Diarization is a 0.1B speaker-diarization model available through supported Badgr execution routes.
Last validated or meaningfully updated: 2026-10-10
✓ Verified by Badgr · PyTorch · Run
Last verified 2026-10-10
badgr run --image vllm/vllm-openai:v0.30.0 --gpu RTX_4090 --funding paid --max-cost 2 --max-runtime 40 --cmd 'pip install nemo-toolkit[asr] (NeMo main) ... SortformerEncLabelModel.from_pretrained("nvidia/Nemotron-3-Diarization").diarize(audio=["meeting.flac"])'- GPU
- NVIDIA RTX A5000Reported by the provider, not confirmed inside the container
- Command runtime
- 320s
- Deployment
- dep-155f7bb71a
- ✓ Workload container started
- ✓ Command exited with code 0
- ✓ The model loaded and diarized a 10-second speech clip
- ✓ Segments with speaker labels were returned
- ✓ Teardown completed
The test clip is a single speaker, so it shows one speaker label across 0.5-10.1 s, not a multi-speaker separation. It needs NeMo from its main branch: the released nemo-toolkit on PyPI fails to build this model (unsupported rope attention). The request asked for an RTX 4090 and the job reported an RTX A5000. The command is abridged and the elapsed time is estimated from the job's cost and hourly rate.
Raw evidence
deployment=dep-155f7bb71a status=succeeded user_command_exit=0 gpu NVIDIA RTX A5000 | 11.5s | segments: 2 0.500 2.370 speaker_0 2.570 10.120 speaker_0
Run from Badgr's local development environment. Documents that this job works end-to-end for this model.
Availability
Not yet Badgr AI API
Not currently validated
✓ Dedicated endpoint
Available through Badgr
✓ Custom GPU deployment
Available through Badgr
Deployment profile
- Architecture
- Nemotron3DiarizationForAudioFrameClassification
- Licence
- OpenMDW-1.1
- Minimum VRAM
- ~5GB
- Context
- Task dependent
Recommended: RTX 4090 24GB. VRAM is an estimate and increases with context, cache, concurrency, and runtime overhead.
Model identity
- Updated
- 16 days ago
- HuggingFace revision
- f667ed73ae
Popularity and trust
Downloads (last month)
72.3K
Likes
752
Spaces using this model
7
Files and formats
Weight formats
Safetensors, GGUF
Repository files
16
Chat template
Not detected
GPU deployment scenarios
Estimated from parameter count and quantisation. Will switch to “Verified by Badgr” once a real run backs a tier.
Testing
RTX 4090 24GB
Short context, low concurrency
Small production
L40S 48GB
8K context, 1-4 concurrent requests
Higher throughput
A100 80GB
32K context, continuous batching
Large production
RTX 4090 24GB
Higher concurrency
Estimated — multimodal inputs, long context, and concurrency all increase real VRAM use beyond this estimate.
Also runs well on L40S 48GB, A100 40GB. Available in United States, Europe.
Serving compatibility
| Runtime | Status | Notes |
|---|---|---|
| vLLM | Estimated | Used for Badgr dedicated endpoints |
| Transformers | Unknown | Basic fallback |
| llama.cpp | Declared by source | GGUF weights available in repo |
| SGLang | Unknown | Not evaluated |
| Ollama | Unknown | Not evaluated |
| Diffusers | Unknown | Not evaluated |
| PyTorch | Verified for Run | Confirmed by a real badgr run job (deployment dep-155f7bb71a); serving is not verified |
| TensorRT-LLM | Unknown | Conversion may be required |
| TGI | Unknown | Not evaluated |
Ready-to-run Badgr configurations
Quick test
badgr serve nvidia/Nemotron-3-Diarization --max-cost 2Validated dedicated endpoint command
badgr serve nvidia/Nemotron-3-Diarization --gpu RTX 4090 --max-cost 10Advanced configuration (defaults to automatic)
- Runtime
- Quantisation
- GPU and GPU count
- Context length
- Maximum concurrency
- Region
- Maximum hourly spend
- Persistent or capped runtime
Estimated pricing
Provider estimate: cold start
2–5 min
Provider estimate: endpoint cost
$0.19 – $0.47/hr
Idle cost
$0 when stopped
Per-token throughput cost is not shown here: Badgr has not benchmarked this model yet, and this page does not display figures it cannot back with real data.
Recommended Badgr routes
RTX 4090 24GB · United States
Availablefrom $0.19/hr
Best for
ComfyUI, Batch Inference, Qwen 2.5 7B Instruct
Startup: 2–5 min
Reliability: Standard
Last checked: just now
RTX 4090 24GB · United States
Availablefrom $0.44/hr
Best for
ComfyUI, Batch Inference, Qwen 2.5 7B Instruct
Startup: 2–5 min
Reliability: Standard
Last checked: just now
RTX 4090 24GB · Europe
Estimatedstarting from $0.45/hr
Badgr estimated starting price
Best for
ComfyUI, Batch Inference, Qwen 2.5 7B Instruct
Startup: 2–5 min
Reliability: Standard
Last checked: Live marketplace price not available yet
RTX 4090 24GB · Europe
Estimatedstarting from $0.45/hr
Badgr estimated starting price
Best for
ComfyUI, Batch Inference, Qwen 2.5 7B Instruct
Startup: 2–5 min
Reliability: Standard
Last checked: Live marketplace price not available yet
RTX 4090 24GB · Europe
Estimatedstarting from $0.45/hr
Badgr estimated starting price
Best for
ComfyUI, Batch Inference, Qwen 2.5 7B Instruct
Startup: 2–5 min
Reliability: Standard
Last checked: Live marketplace price not available yet
RTX 4090 24GB · United States
Availablefrom $0.47/hr
Best for
ComfyUI, Batch Inference, Qwen 2.5 7B Instruct
Startup: 2–5 min
Reliability: Standard
Last checked: just now
Known limitations
- Memory use increases with context length and concurrency.