It works on a range of documents (see usage and benchmarks).
surya-ocr-2.5
datalab-to·Instruct·released 25 Oct 2026
ollama run hf.co/datalab-to/surya-ocr-2.5-gguf:Q8_0We also ship smaller models for line-level text detection and ocr error detection. It works on a range of documents (see usage and benchmarks).
How much VRAM does surya-ocr-2.5 need?
At FP16 / BF16 surya-ocr-2.5 needs about 2.0 GB of GPU memory.
| Quantisation | Total |
|---|---|
| 4-bitRecommended default for local use | ≈1.0 GB |
| 8-bitLossless in practice, twice the size of Q4 | ≈1.2 GB |
| FP16 / BF16recommendedFull precision, as released | ≈2.0 GB |
Estimates for the weights plus runtime overhead; activations vary with resolution and batch size. Pick a GPU above and the fit column checks every row against it. Weights marked “file” use the real GGUF download size; “est.” rows are computed from the parameter count.
Run surya-ocr-2.5 locally
--quantQ8_0default; pick your GPU to tailor itdatalab-to/surya-ocr-2.5-gguf
ollama run hf.co/datalab-to/surya-ocr-2.5-gguf:Q8_0 llama-server -hf datalab-to/surya-ocr-2.5-gguf:Q8_0Downloads the GGUF on first run. Installed through llama.app? The same flags work with `llama serve`.
lms get datalab-to/surya-ocr-2.5-gguf@Q8_0Then load it from the LM Studio app or with `lms load`.
vllm serve datalab-to/surya-ocr-2.5services:
vllm:
image: vllm/vllm-openai:latest
command: datalab-to/surya-ocr-2.5
ports:
- "8000:8000"
ipc: host
volumes:
- ~/.cache/huggingface:/root/.cache/huggingface
deploy:
resources:
reservations:
devices:
- driver: nvidia
count: all
capabilities: [gpu]API on port 8000.
sglang serve datalab-to/surya-ocr-2.5 --host 0.0.0.0 --port 30000services:
sglang:
image: lmsysorg/sglang:latest
command: sglang serve datalab-to/surya-ocr-2.5 --host 0.0.0.0 --port 30000
ports:
- "30000:30000"
ipc: host
volumes:
- ~/.cache/huggingface:/root/.cache/huggingface
deploy:
resources:
reservations:
devices:
- driver: nvidia
count: all
capabilities: [gpu]API on port 30000.
from transformers import pipeline
pipe = pipeline("image-text-to-text", model="datalab-to/surya-ocr-2.5", dtype="auto", device_map="auto")
messages = [
{
"role": "user",
"content": [
{"type": "image", "url": "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/bee.jpg"},
{"type": "text", "text": "Describe this image."},
],
}
]
out = pipe(text=messages, max_new_tokens=256)
print(out[0]["generated_text"][-1]["content"])Quantised versions and downloads
| Repository | Format | Quant | Size |
|---|---|---|---|
| datalab-to/surya-ocr-2.5-gguf | GGUF | Q8_0 | 0.6 GB |
| fwgpiyawudk/surya-ocr-2.5-mlx-4bit | MLX | 4BIT | — |
| kltng/surya-ocr-2.5-mlx-8bit | MLX | 8BIT | — |
| fwgpiyawudk/surya-ocr-2.5-mlx-8bit | MLX | 8BIT | — |
About surya-ocr-2.5
We also ship smaller models for line-level text detection and ocr error detection. It works on a range of documents (see usage and benchmarks). Our managed platform runs both Surya, and variants of our highest accuracy model, Chandra. Get started with $5 in free credits — sign up (takes under 30 seconds) or try our free public playground. Surya is named for the Hindu sun god, who has universal vision. Each row links to four annotated views of the same page: text-line detection, OCR, layout and reading order (Surya 2.5). The example pages are public pages from OmniParseBench. The newspaper (The Alpha, 1881), reference book and math paper are from olmOCR-bench (Ai2, ODC-BY 1.0). The government form is a US federal form (public domain). The handwritten letter is from the Library of Congress, Manuscript Division, Carrie Chapman Catt Papers (public domain). The corporate page is from ParseBench (Apache 2.0). The Japanese form is made from a form of the National Tax Agency, Japan, under the Public Data License 1.0: 「国税庁の様式」(国税庁)(https://www.nta.go.jp/)を加工して作成. The OCR column renders Surya 2.5's HTML output; layout and reading order are drawn from its blocks, and the detection image is from the text-line detector.
Full model card on Hugging Face →Alternatives to surya-ocr-2.5
All alternatives →OvisOCR3 is a unified 0.8B multimodal model for OCR and document understanding. It integrates document parsing, general OCR, key information extraction, document figure captioning, and cross-page table reconstruction into a single compact model.
Best OCR model . LightOnOCR-2-1B is LightOn's flagship OCR model, refined with RLVR training for maximum accuracy. We recommend this variant for most OCR tasks.
Smallest model. LightOnOCR-3-0.8B is the lightest model of LightOn's LightOnOCR-3 family, with the same layout and visual understanding features.
We are pleased to announce the release of OvisOCR2, a compact 0.8B end-to-end model for page-level document parsing. Given a document page image, OvisOCR2 generates a Markdown representation in natural reading order, covering text, formulas, tables, and visual regions.
PaddleOCR-VL-1.6: Expanding the Frontier of Document Parsing with Under-Optimized Region Refinement and Progressive Post-Training
surya-ocr-2.5 FAQ
- How much VRAM does surya-ocr-2.5 need?
- About 2.0 GB at FP16 / BF16.
- Can surya-ocr-2.5 run on the RTX 4060?
- Yes. It fits at FP16 / BF16 (about 2.0 GB). It fits comfortably on the Apple M4 (16 GB), Radeon RX 6700 XT, RTX 3060 12GB.
- Can I use surya-ocr-2.5 commercially?
- Yes, with conditions. It is released under OpenRAIL. Commercial use allowed, but with use-based restrictions that pass on to derivatives.
- How do I run surya-ocr-2.5 with Ollama or llama.cpp?
- With Ollama: ollama run hf.co/datalab-to/surya-ocr-2.5-gguf:Q8_0. With llama.cpp: llama-server -hf datalab-to/surya-ocr-2.5-gguf:Q8_0, which starts an OpenAI-compatible server.
Rate this model
Data from Hugging Face, the model card and public leaderboards. Memory and speed figures are estimates; real numbers depend on the runtime, context length and drivers. Always check the licence yourself before commercial use.
0 comments