NVIDIA Nemotron Parse 2.0 — Hardware Requirements & GPU Compatibility
Vision
NVIDIA Nemotron Parse 2.0 is a sub-1-billion-parameter vision-encoder-decoder model purpose-built for document parsing rather than open-ended chat: given a page image, it outputs structured text with layout classes, bounding boxes, and reading order for elements like titles, paragraphs, tables, charts, and footnotes. It pairs a ViT-H vision encoder based on NVIDIA's C-RADIO with a 10-block mBART decoder, and over its predecessor v1.2 it adds roughly 20,000 new vocabulary tokens for more efficient multilingual (especially CJK and Indic-script) OCR, a dedicated chart class for chart-to-table parsing, and stronger table detection and text extraction. At under a billion parameters it runs on a single modest GPU. License is the OpenMDW License Agreement version 1.1, permitting both commercial and non-commercial use; the bundled tokenizer is separately licensed under CC-BY-4.0. It was published in August 2026.
Runs great on a 24 GB GPU
Runs from 8 GB (BF16 · 32K)Best on 24 GB (BF16 · ~329 tok/s · 32K)Macs from 8 GB (BF16 · ~36 tok/s)
openmdw-1.1
Reference GPU · RTX 4090 24 GB
Estimates, ±20% · how we estimate
Run NVIDIA Nemotron Parse 2.0
One recommended way. Every other method, OS and quant is in the step-by-step guide.Each way opens a step-by-step guide for your OS and quant.
Recommended
vllm serve nvidia/NVIDIA-Nemotron-Parse-2.0 --max-model-len 16384vLLM repo with its own quantisation
Download: Weights on Hugging Face
Ways to runOther ways
each opens step-by-stepRun in cloud
Runs on RTX 3060 12GB (8 GB headroom) · BF16
- Generation speed
- ~118 tok/s
- generation speed
- Cost per 1M output tokens
- $0.14
- per 1M output tokens
How Much VRAM Does NVIDIA Nemotron Parse 2.0 Need?
Select a quantization level; the listed hardware adheres to it.
| Quantization | VRAM | +32K | On RTX 4090 (reference) | |
|---|---|---|---|---|
| 2.0 GB | 5 GB | 16-bit · bfloat — Brain floating point 16 — preferred for training | Fits: 32K context · ~329 tok/s |
Selected BF16
Brain floating point 163.4GBat 16K context
- Weights
- 1.5 GB
- KV cache
- 1.6 GB
- Overhead
- 0.3 GB
Fits RTX 4090 (reference) · 20.6 GB free · ~329 tok/s
≥50 tok/s needs ≥153 GB/s · RTX 5090 ~585, RTX 3090 Ti ~329
Which Hardware Can Run NVIDIA Nemotron Parse 2.0?
BF16 · 2.0 GBNVIDIA Nemotron Parse 2.0 (BF16) requires 2.0 GB of VRAM to load the model weights. For comfortable inference with headroom for KV cache and system overhead, 3+ GB is recommended. 52 GPUs can run it, including NVIDIA GeForce RTX 5090, NVIDIA GeForce RTX 3090 Ti. 163 devices with unified memory can run NVIDIA Nemotron Parse 2.0, including Mac Studio (M3 Ultra, 256GB), Mac Studio (M3 Ultra, 512GB).
≤ 11 GB
≤ 11 GB tier →fits · 4.6 GB free · up to 32K ctx
16 GPUs · 1 Mac class · 2 device families
RTX 3080~248Simulate output on RTX 3080RTX 2080 Ti~201Simulate output on RTX 2080 TiRTX 3070 Ti~199Simulate output on RTX 3070 TiM3*8 GB~36Simulate output on iMac 24" M3 (8 GB)Orin Nano~33Simulate output on NVIDIA Jetson Orin Nano 8GB (Super)iPhone~24Simulate output on iPhone 1720 GB
fits · 16.6 GB free · up to 32K ctx
1 GPU
RX 7900 XT~241Simulate output on RX 7900 XT32–40 GB
fits · 28.6 GB free · up to 32K ctx
1 GPU · 3 Mac classes · 4 device families
RTX 5090~585Simulate output on RTX 5090M5 Max36 GB~162Simulate output on Mac Studio M5 Max (36 GB)M532 GB~54Simulate output on MacBook Air 13" M5 (32 GB)Orin~67Simulate output on NVIDIA Jetson AGX Orin 32GBSnapdragon X Elite (X1E-80-100)~44Simulate output on Snapdragon X Elite Copilot+ PC
By VRAM tier · typical GPUs and Macs at BF16
Alternatives to NVIDIA Nemotron Parse 2.0
Nemotron family · 45 models →Similar models
6| Model | Score | VRAM | tok/s | Why it is similar |
|---|---|---|---|---|
| LightOnOCR 3 0.8Blightonai | — | 0.9 GB | ~762 | same size |
| PaddleOCR VL 1.6PaddlePaddle | — | 1 GB | ~736 | same size |
| Qwen3.5 0.8BAlibaba | — | 0.9 GB | ~753 | same size |
| InternVL3 1BOpenGVLab | — | 2 GB | ~1057 | same size |
| InternVL2 1BOpenGVLab | — | 2 GB | ~1057 | same size |
| MobileLLM R1.5 950MMeta | — | 3.5 GB | ~314 | same size |
Specifications
- Publisher
- NVIDIA
- Family
- Nemotron
- Parameters
- 903M
- Architecture
- NemotronParseForConditionalGeneration
- Vocabulary Size
- 72,256
- Release Date
- 2026-06-30
- License
- openmdw-1.1
Frequently Asked Questions
How much VRAM does NVIDIA Nemotron Parse 2.0 need?
NVIDIA Nemotron Parse 2.0 requires 2.0 GB of VRAM at BF16.
VRAM = Weights + KV Cache + Overhead
Weights = 903M × 16 bits ÷ 8 = 1.8 GB
KV Cache + Overhead ≈ 0.2 GB (at 2K context + ~0.3 GB framework)
Fit ratings and hardware model lists check this model with room for a 16K-token context, which needs a little more memory.
VRAM usage by quantization
Can I run NVIDIA Nemotron Parse 2.0 on a Mac?
Yes — MacBook Air 13" M3 (8 GB) and 145 other Macs can run NVIDIA Nemotron Parse 2.0. Apple Silicon uses unified memory, so the model shares RAM with the system. At BF16 you need at least 2.0 GB of usable unified memory (RAM minus macOS overhead).
Can I run NVIDIA Nemotron Parse 2.0 locally?
Yes — NVIDIA Nemotron Parse 2.0 can run locally on consumer hardware. Popular tools include Ollama, LM Studio, and llama.cpp.
How fast is NVIDIA Nemotron Parse 2.0?
At BF16, NVIDIA Nemotron Parse 2.0 can reach ~2412 tok/s on AMD Instinct MI350X. On NVIDIA GeForce RTX 4090: ~329 tok/s. Speed depends mainly on GPU memory bandwidth. Real-world results typically within ±20%.
tok/s = (bandwidth GB/s ÷ model GB) × efficiency
Example: NVIDIA B200 → 8000 ÷ 1.99 × 0.65 = ~2613 tok/s
Estimated speed at BF16 (2.0 GB)
Real-world results typically within ±20%. Speed depends on batch size, quantization kernel, and software stack.
What's the download size of NVIDIA Nemotron Parse 2.0?
At BF16, the download is about 1.81 GB.
Which GPUs can run NVIDIA Nemotron Parse 2.0?
52 consumer GPUs can run NVIDIA Nemotron Parse 2.0 at BF16. Top options include AMD Radeon RX 6700 XT, AMD Radeon RX 6800, AMD Radeon RX 6800 XT. 52 GPUs have plenty of headroom for comfortable inference.
Which devices can run NVIDIA Nemotron Parse 2.0?
163 devices with unified memory can run NVIDIA Nemotron Parse 2.0 at BF16 (2.0 GB), including AMD Ryzen AI 9 HX 370 (Strix Point) Laptop, ASUS Ascent GX10, Beelink GTR9 Pro (Ryzen AI Max+ 395, 128 GB), Framework Desktop (Ryzen AI Max+ 395, 128 GB). Apple Silicon Macs use unified memory shared between CPU and GPU, making them well-suited for local LLM inference.