SmolDocling vs InternVL3.5
Pick SmolDocling only for small English experiments where a 256M model and DocTags output matter, and note that it is a preview release superseded by Granite-Docling. Pick InternVL3.5 for document question answering and for teams that want one open model family at several sizes, from 1B to 241B-A28B. The main split is that SmolDocling is a narrow page converter with layout tags, while InternVL3.5 is a general vision language model with no bounding boxes by default.
SmolDoclinghuggingface/smoldocling | InternVL3.5opengvlab/internvl | |
|---|---|---|
| Type | Open source | Vision model |
| Price | 1 credit / page | 2 credits / page |
| Per 1,000 pages | $1.00 | $2.00 |
| Typical speed | 1.0 s / page | 3.0 s / page |
| Best for | Very cheap page conversion | Self-hosted document understanding |
| Inputs | PDF, images | images |
| Languages | Mainly English | Multilingual |
| Max pages per request | 500 | 500 |
| Tables | ✓ Yes | ✓ Yes |
| Handwriting | No | No |
| Math | ✓ Yes | ✓ Yes |
| Bounding boxes | ✓ Yes | No |
| Confidence | No | No |
| Forms | No | No |
| Multilingual | No | ✓ Yes |
| On OCRRouter | Available | Available |
Prices are what you pay on OCRRouter. Speed is the provider's typical figure, not a benchmark we ran.
Which to choose
Choose SmolDocling if
- It outputs DocTags, Markdown, HTML, JSON, LaTeX and bounding boxes, so you get layout and location data for each element.
- At 256M parameters it runs on modest hardware, and the authors report about 0.35 seconds per page on an A100 with vLLM.
- One model covers text, tables (OTSL), code, formulas and charts with location tags, for English page conversion on a low budget.
Choose InternVL3.5 if
- Nine sizes from 1B to 241B let you trade accuracy for cost, and the MoE variants 20B-A4B and 30B-A3B activate few parameters per token.
- It suits document question answering and self-hosted multimodal pipelines, since it is a general vision language model.
- Apache-2.0 licence, and models up to 30B fit on a single A100 per the model card.
Key differences
| SmolDocling | InternVL3.5 | |
|---|---|---|
| Model type | Document conversion VLM that outputs DocTags for Docling | General open-weight vision language model family from OpenGVLab |
| Size | 256M parameters | 1B to 241B-A28B; the 8B variant is 8.5B |
| Licence | Apache-2.0 per the model summary (the metadata tag says CDLA-Permissive-2.0) | Apache-2.0 |
| Output formats | DocTags, Markdown, HTML, JSON, LaTeX, bounding boxes | Plain text, Markdown, JSON; no bounding boxes or layout output by default |
| Benchmarks | TEDS structure on FinTabNet 0.82, full-page OCR F1 0.80, OCRBench 338 | Not published as text; the model card shows OCR scores only as chart images |
| Languages | English only | Not published |
| Known failure or status | Preview release. Output can loop on some images (Docling issue #1515) and the paper lists endless token repetition and missing closing tags. Granite-Docling scores higher, e.g. TEDS 0.97 vs 0.82. | The largest 241B-A28B model needs eight A100 GPUs, and the 38B model needs two A100s with tensor parallelism |
Sources
- SmolDocling model card
- SmolDocling paper
- Docling issue #1515
- Granite-Docling comparison table
- Hugging Face model card (8B)
- InternVL GitHub
What each one is
SmolDocling
Tiny 256M vision model from Hugging Face and IBM that outputs DocTags. Predecessor of Granite Docling.
SmolDocling detailsInternVL3.5
Shanghai AI Lab's open vision-language family (1B to 241B, August 2025), strong on OCR and document benchmarks.
InternVL3.5 detailsTry both on your own file
Run the file once with each model and compare the output.
Use one, fall back to the other
{"model": "huggingface/smoldocling","fallbacks": ["opengvlab/internvl"],"file_url": "https://example.com/document.pdf"}Questions
More comparisons
SmolDocling alternativesInternVL3.5 alternativesBest OCR models in 2026