Nemotron 3 Nano 30B A3B vs Baidu Unlimited-OCR-2

Nemotron 3 Nano 30B A3B

A fast hybrid Mamba-Transformer cousin with 3.5B active parameters for reasoning and tool calling over very long inputs, in every major local runtime. A 4-bit build needs about 22 to 23 GB. The main risks are that NVIDIA's documented vLLM path runs repository code, no safety results are published and NVIDIA hasn't answered reports since March.

Baidu Unlimited-OCR-2

A small MIT document-parsing cousin that reads up to hundreds of pages in one request and returns ordered layout blocks with boxes and HTML tables. It needs an NVIDIA GPU with about 12 GB, through Baidu's remote code or its own SGLang build. The risks are running the publisher's code and accuracy figures that are Baidu's own.

Nemotron 3 Nano 30B A3B

1. Start it with ollama run nemotron-3-nano:30b (18 GB MLX on a Mac, 24 GB q4_K_M elsewhere)
2. On the reference machine use BF16 with vLLM, after wget https://huggingface.co/nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B-BF16/resolve/main/nano_v3_reasoning_parser.py, then vllm serve nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B-BF16 --max-model-len 262144 --trust-remote-code --enable-auto-tool-choice --tool-call-parser qwen3_coder --reasoning-parser-plugin nano_v3_reasoning_parser.py --reasoning-parser nano_v3
3. NVFP4 needs an NVIDIA Blackwell GPU, and FP8 isn't documented on gfx1151
4. Without the qwen3_coder tool parser and the nano_v3 reasoning parser, tool calls and reasoning come back as plain text
5. Card sampling is temperature 0.6 and top_p 0.95 for tool calling, temperature 1.0 and top_p 1.0 for reasoning
6. Thinking is on by default, so pass enable_thinking as false in chat_template_kwargs for short tasks
7. Set VLLM_ALLOW_LONG_MAX_MODEL_LEN=1 and raise --max-model-len only when you need more than 262,144 tokens
8. In SGLang the main branch names the reasoning parser nemotron_3, not the card's nano_v3
9. Filter untrusted tool results before they reach the model, as NVIDIA warns prompt injection can lead to code execution
10. Validate JSON output yourself or use the runtime's guided decoding, since none is documented for the model

Baidu Unlimited-OCR-2

1. Quickest local run, pip install torch==2.13.0 torchvision pillow transformers==4.52.1 huggingface-hub==0.36.0 tokenizers==0.21.4 addict==2.4.0 einops==0.8.2 easydict==1.13, then AutoModel.from_pretrained("baidu/Unlimited-OCR-2", trust_remote_code=True, use_safetensors=True, torch_dtype=torch.bfloat16).eval().cuda() and model.infer_multi(tokenizer, image_files=pages, output_path="out")
2. To serve it, download the repository, install wheel/sglang-0.5.20.dev10+g8d62f5fb9c-py3-none-any.whl with pymupdf and fastokens in a Python 3.10 venv, and start python -m sglang.launch_server --model-path baidu/Unlimited-OCR-2 --chat-template unlimited-ocr --enable-custom-logit-processor --enable-streaming-session --disable-radix-cache --mem-fraction-static 0.7 --max-prefill-tokens 2048 with the SGLANG_UOCR_ variables from the card
3. Send all pages of a document in one request, images first, then the text <PAGE><image> repeated per page and Multi page parsing., with temperature 0, skip_special_tokens false, max_tokens of about 8,000 per page and the DeepseekOCRNoRepeatNGramLogitProcessor (ngram_size 35, window_size 1024 for several pages)
4. The output is layout blocks, not Markdown; run utils/process_unlimitedOCR2.py from github.com/baidu/Unlimited-OCR, or convert(raw_output) in Python, to get Markdown with HTML tables
5. Render PDFs to images first (the card uses PyMuPDF at 300 DPI), keep save_results off for untrusted documents, and treat extracted text as untrusted before passing it to another model