What GPU do I need to run nanonets/Nanonets-OCR-s?
A 3.8B vision-language model that reads images alongside text. 3.8B parameters, published in BF16. View on Hugging Face
Nanonets-OCR-s is published by nanonets on Hugging Face, with 349,113 downloads and 1,593 likes to date. It's a Qwen2_5_VLForConditionalGeneration model built for image-text-to-text, published natively in BF16.
What Nanonets-OCR-s is
Nanonets-OCR-s is a 3.8B-parameter vision-language model published by Nanonets on Hugging Face. It is released under Not stated.
License note: no license tag published on the model's Hugging Face card; check the repo directly before any commercial use. Facts in this section are sourced from Nanonets-OCR-s's Hugging Face model card, not benchmarked by Aquanode.
What it's used for
- Image understanding and captioning
- Visual question answering
- Document/OCR-style reading
VRAM required & cheapest live GPU fit
Required VRAM = weight size at each precision times a flat 1.2 overhead for activations and fragmentation. The KV-cache grows with context and is not in that factor; it is listed per model below where the architecture is published. Full formula and assumptions: methodology.
| Precision | Weight size | Required VRAM | Cheapest live fit | GPUs needed | Est. $/hr (full fit) |
|---|---|---|---|---|---|
| BF16 | 7.0 GB | 8.4 GB | RTX 3060 | 1 | $0.110/hr |
| FP8 (quantized) | 3.5 GB | 4.2 GB | RTX 5060 Ti | 1 | $0.165/hr |
| INT4 (quantized) | 1.7 GB | 2.1 GB | RTX 3070 | 1 | $0.088/hr |
A GPU is only matched to a row if its hardware supports that precision, and the primary recommendation is always a single-GPU fit when one exists.
INT4 requires a quantized checkpoint actually published for this model, check its Hugging Face page before relying on this row.
Cheapest way to run Nanonets-OCR-s at its published (BF16) precision: 1× RTX 3060, at $0.110/hr per GPU ($0.110/hr total). Quantizing to FP8 or INT4 (rows above) can cost less, but requires a compatible quantized checkpoint to exist for this model.
Nanonets-OCR-s: KV cache by context length
The KV cache is the memory the attention layers hold for every token of context, on top of the weights. It is not part of the flat 1.2x overhead in the table above, grows with context length and with every concurrent request, and is why a long-context deployment needs more VRAM than the table shows.
Grouped-query attention: every layer caches keys and values for a few shared KV heads. Cached per token: 36,864 bytes at 16-bit.
| Context | KV cache, one sequence |
|---|---|
| 4K | 0.14 GB |
| 32K | 1.1 GB |
| 125K (model maximum) | 4.4 GB |
Computed from the layer, head and window counts in the model's own configuration (nanonets/Nanonets-OCR-s), at 16 bits per cached value. Each concurrent sequence needs its own cache, on top of the weights.
Nanonets-OCR-s: common questions
Does Nanonets-OCR-s fit on a 12 GB GPU?
Yes. At BF16 it needs 8.4 GB of VRAM, so a 12 GB card holds it with 3.6 GB to spare. An 8 GB card is not enough for it at BF16.
What is the least VRAM Nanonets-OCR-s can run in?
2.1 GB, at INT4 (quantized), which fits a 6 GB card, against 8.4 GB at BF16. That row assumes an INT4 checkpoint has actually been published for this model, so check its Hugging Face page before planning around it.
Does quantizing Nanonets-OCR-s lower the GPU bill?
Yes. At BF16 the cheapest live fit is one RTX 3060 at $0.110/hr. At INT4 (quantized) it drops to one RTX 3070 at $0.088/hr, provided a quantized checkpoint exists for it.
How to run Nanonets-OCR-s
Run Nanonets-OCR-s with vLLM
From nanonets/Nanonets-OCR-s's own deployment docs.
vllm serve nanonets/Nanonets-OCR-sSource: https://huggingface.co/nanonets/Nanonets-OCR-s/raw/main/README.md
Run Nanonets-OCR-s with GGUF quantizations
Prebuilt GGUF weights published at unsloth/Nanonets-OCR-s-GGUF. Run with llama.cpp's llama-server or load the repo directly in LM Studio.
llama-server -hf unsloth/Nanonets-OCR-s-GGUFDeploy Nanonets-OCR-s on Aquanode
Aquanode has no one-click deploy template for Nanonets-OCR-s; you install the inference engine yourself with the commands below. Aquanode sells GPU pods billed per second, not a hosted inference API.
- Launch a bare GPU pod sized to the requirement above (1× RTX 3060 or larger).
- Open a terminal on the pod, or save one of the commands above as a startup script so it runs automatically the first time the pod boots.
- Run the command and connect to the resulting endpoint.
Weight-to-VRAM math, the fit rules, and how live prices are normalized: full methodology.
More Qwen2.5 models
- Qwen2.5-VL-3B-Instruct (3.8B, BF16)
- Qwen2.5-VL-7B-Instruct (8.3B, BF16)
- NuMarkdown-8B-Thinking (8.3B, BF16)
- Qwen2.5-VL-32B-Instruct (33.5B, BF16)
- Qwen2.5-VL-72B-Instruct (73.4B, BF16)
Lineage
Based on Qwen2.5-VL-3B-Instruct.
Alternatives at this size
Other models for image-text-to-text within about a third of Nanonets-OCR-s's 3.8B parameters, from other model lines.
- Unlimited-OCR (3.3B, BF16)
- chandra-ocr-2 (5.3B, BF16)
- DeepSeek-OCR (3.3B, BF16)
- gemma-3-4b-it (4.3B, BF16)
- Phi-3.5-vision-instruct (4.1B, BF16)
More on Nanonets-OCR-s
Fits on a 12 GB GPU at BF16: every model that fits in 12 GB.
Fits on an 8 GB GPU at FP8: every model that fits in 8 GB.
Related reading: H100 pricing and specs, How much VRAM you need for LLMs, Serving LLMs with vLLM, vLLM vs TensorRT-LLM vs SGLang, Best GPU for LLM inference, and What AI model fine-tuning is.
Your own GPU cloud accounts
One console for the GPU clouds you already rent from. Your keys stay on your machine.
$20 a month, flat