Nemotron 3 models

19 Nemotron 3 models from NVIDIA on Hugging Face, from 631M to 560.5B parameters, published by NVIDIA in F32 and BF16, with contexts from 256K to 1M tokens. The smallest official model, nemotron-3.5-asr-streaming-0.6b, needs about 2.9 GB of VRAM at its published precision; the cheapest live fit is Quadro M4000 at $0.046/hr.

Part of the Nemotron series

Pick a size

One row per official Nemotron 3 size: the VRAM it needs at each precision, the cheapest GPU that holds it today, and the KV cache for a 32K-token context.

ModelParametersNative VRAMFP8 VRAMINT4 VRAMLive GPU fit (native)Est. $/hrKV cache at 32K
nemotron-3.5-asr-streaming-0.6b638M2.9 GB0.7 GB0.4 GBQuadro M4000$0.046/hrnot published for this architecture
NVIDIA-Nemotron-3-Nano-4B-BF164.0B8.9 GB4.4 GB2.2 GBRTX 5060 Ti$0.145/hr0.50 GB
NVIDIA-Nemotron-3-Nano-30B-A3B-BF1631.6B70.6 GB35.3 GB17.6 GBA100$1.31/hr0.19 GB
NVIDIA-Nemotron-3.5-Transcribe-30B-A4B33.7B75.3 GB37.7 GB18.8 GBA100$1.31/hrnot published for this architecture
NVIDIA-Nemotron-3-Super-120B-A12B-BF16123.6B276 GB138 GB69.1 GBRTX A6000 × 6$2.18/hr0.25 GB
NVIDIA-Nemotron-3-Ultra-550B-A55B-BF16560.5B1253 GB626 GB313 GBAMD MI300X × 7$18.40/hrnot published for this architecture

VRAM is weights times a flat 1.2 overhead; the KV cache is a separate per-model figure at 16-bit, shown where the architecture is published. See the methodology.

Official models (10)

ModelParametersPublished asVRAM neededLive GPU fitEst. $/hr
nemotron-3.5-asr-streaming-0.6b638MF322.9 GBQuadro M4000$0.046/hr
NVIDIA-Nemotron-3-Nano-4B-BF164.0BBF168.9 GBRTX 5060 Ti$0.145/hr
NVIDIA-Nemotron-3-Nano-30B-A3B-Base-BF1631.6BBF1670.6 GBA100$1.31/hr
NVIDIA-Nemotron-3-Nano-30B-A3B-BF1631.6BBF1670.6 GBA100$1.31/hr
NVIDIA-Nemotron-3.5-Lightning-30B-A3B-Base-BF1631.6BBF1670.6 GBA100$1.31/hr
NVIDIA-Nemotron-3.5-Lightning-30B-A3B-BF1631.6BBF1670.6 GBA100$1.31/hr
NVIDIA-Nemotron-3.5-Transcribe-30B-A4B33.7BBF1675.3 GBA100$1.31/hr
NVIDIA-Nemotron-3-Super-120B-A12B-Base-BF16123.6BBF16276 GBRTX A6000 × 6$2.18/hr
NVIDIA-Nemotron-3-Super-120B-A12B-BF16123.6BBF16276 GBRTX A6000 × 6$2.18/hr
NVIDIA-Nemotron-3-Ultra-550B-A55B-BF16560.5BBF161253 GBAMD MI300X × 7$18.40/hr

Quantized versions (5)

ModelParametersPublished asVRAM neededLive GPU fitEst. $/hr
NVIDIA-Nemotron-3.5-Lightning-30B-A3B-NVFP4-DSpark764MBF161.7 GBRTX 3070$0.088/hr
NVIDIA-Nemotron-3-Nano-4B-FP84.0BF8_E4M34.4 GBRTX 5060 Ti$0.145/hr
NVIDIA-Nemotron-3-Nano-30B-A3B-FP831.6BF8_E4M335.3 GBL40$0.759/hr
NVIDIA-Nemotron-3-Super-120B-A12B-FP8123.6BF8_E4M3138 GBAMD MI300X$2.63/hr
NVIDIA-Nemotron-3-Ultra-550B-A55B-FP8-block560.5BF8_E4M3626 GBRTX PRO 6000 × 7$9.63/hr

Fine-tunes and community models (4)

ModelParametersPublished asVRAM neededLive GPU fitEst. $/hr
OpenThai2.0-OCR-Markdown631MF322.8 GBQuadro M4000$0.046/hr
nemotron3-8b8.5BBF1619.1 GBRTX A5000$0.176/hr
Fastino-Nemotron-3.5-Lightning-Healthcare31.6BBF1670.6 GBA100$1.31/hr
phonellm-alpha-131.6BBF1670.6 GBA100$1.31/hr

VRAM is for the precision the model is published in: the weights times a flat 1.2 overhead, with the KV cache not included. See the methodology. The fit shown is the lowest-priced single GPU type that holds the model at that precision, or the lowest-priced multi-GPU set (up to 8) when none does.

Run and fine-tune Nemotron 3

More from NVIDIA

Start ComfyUI on a rented GPU.

© 2026 Aquanode. All rights reserved.

All trademarks, logos and brand names are the property of their respective owners.