SAVRN Model Hub · Comparisons
DeepSeek-OCR vs dots.ocr
DeepSeek-OCR has 3.3B parameters and dots.ocr has 3B parameters; both are released under MIT License; at 16-bit, DeepSeek-OCR needs about 8 GB (1x MI300X from $1.85 an hour) and dots.ocr about 7.3 GB (1x MI300X from $1.85 an hour).
| Field | DeepSeek-OCR deepseek-ai/DeepSeek-OCR | dots.ocr dots-studio/dots.ocr |
|---|---|---|
| Publisher | DeepSeek | Dots Studio |
| Task | Image and text to text | Image and text to text |
| Modality | Image and text | Image and text |
| Parameters, as reported | 3.3B parameters | 3B parameters |
| Architecture | DeepseekOCRForCausalLM | DotsOCRForCausalLM |
| Library | transformers | dots_ocr |
| Context length | 8,192 tokens | 131,072 tokens |
| Repository size | 6.7 GB | 6.1 GB |
| Artifact formats | safetensors | safetensors |
| License | mit | mit |
| Access | Open weights, no gate | Open weights, no gate |
| Memory at 16-bit (weights and margin) | 8 GB | 7.3 GB |
| Cheapest GPUs at 16-bit, per hour | 1x MI300X, $1.85 | 1x MI300X, $1.85 |
| Memory at 4-bit (weights and margin) | 2 GB | 1.8 GB |
| Cheapest GPUs at 4-bit, per hour | 1x MI300X, $1.85 | 1x MI300X, $1.85 |
| Revision viewed | 9f30c71f441d | c0111ce6bc07 |
| Downloads reported by the hub | 2.3M | 710.9k |
| Last observed | 2026-09-20 | 2026-09-18 |
An evaluation row appears only where at least two of these models report the same benchmark with the same stated configuration, metric, unit and setup. Different evaluators stay named in each cell. Values are shown as reported: no unit conversion, no ranking.
Other Reported Results
These results are listed for each model on its own, because the conditions needed to compare them are not stated or do not match. Two results that leave a condition blank are not assumed to share it.
DeepSeek-OCR
| Benchmark | Conditions | Result | Reported by | Revision | Date |
|---|---|---|---|---|---|
| Delores-Lin/MDPBench | Task arMetric arComparison conditions not established | 56.9 | MDPBench leaderboard Reported by a third party |
Evaluated revision not stated | 2026-07-03 |
| Delores-Lin/MDPBench | Task deMetric deComparison conditions not established | 55 | MDPBench leaderboard Reported by a third party |
Evaluated revision not stated | 2026-07-03 |
| Delores-Lin/MDPBench | Task digitalMetric digitalComparison conditions not established | 80.7 | MDPBench leaderboard Reported by a third party |
Evaluated revision not stated | 2026-07-03 |
| Delores-Lin/MDPBench | Task enMetric enComparison conditions not established | 58.3 | MDPBench leaderboard Reported by a third party |
Evaluated revision not stated | 2026-07-03 |
| Delores-Lin/MDPBench | Task esMetric esComparison conditions not established | 44.1 | MDPBench leaderboard Reported by a third party |
Evaluated revision not stated | 2026-07-03 |
| Delores-Lin/MDPBench | Task frMetric frComparison conditions not established | 43.2 | MDPBench leaderboard Reported by a third party |
Evaluated revision not stated | 2026-07-03 |
| Delores-Lin/MDPBench | Task hiMetric hiComparison conditions not established | 52.2 | MDPBench leaderboard Reported by a third party |
Evaluated revision not stated | 2026-07-03 |
| Delores-Lin/MDPBench | Task idMetric idComparison conditions not established | 60.9 | MDPBench leaderboard Reported by a third party |
Evaluated revision not stated | 2026-07-03 |
| Delores-Lin/MDPBench | Task itMetric itComparison conditions not established | 69.3 | MDPBench leaderboard Reported by a third party |
Evaluated revision not stated | 2026-07-03 |
| Delores-Lin/MDPBench | Task jpMetric jpComparison conditions not established | 49.1 | MDPBench leaderboard Reported by a third party |
Evaluated revision not stated | 2026-07-03 |
| Delores-Lin/MDPBench | Task koMetric koComparison conditions not established | 28.2 | MDPBench leaderboard Reported by a third party |
Evaluated revision not stated | 2026-07-03 |
| Delores-Lin/MDPBench | Task latinMetric latinComparison conditions not established | 54.5 | MDPBench leaderboard Reported by a third party |
Evaluated revision not stated | 2026-07-03 |
| Delores-Lin/MDPBench | Task nlMetric nlComparison conditions not established | 52.4 | MDPBench leaderboard Reported by a third party |
Evaluated revision not stated | 2026-07-03 |
| Delores-Lin/MDPBench | Task non_latinMetric non_latinComparison conditions not established | 48.9 | MDPBench leaderboard Reported by a third party |
Evaluated revision not stated | 2026-07-03 |
| Delores-Lin/MDPBench | Task overallMetric overallComparison conditions not established | 51.8 | MDPBench leaderboard Reported by a third party |
Evaluated revision not stated | 2026-07-03 |
| Delores-Lin/MDPBench | Task photographedMetric photographedComparison conditions not established | 42.2 | MDPBench leaderboard Reported by a third party |
Evaluated revision not stated | 2026-07-03 |
| Delores-Lin/MDPBench | Task privateMetric privateComparison conditions not established | 54.5 | MDPBench leaderboard Reported by a third party |
Evaluated revision not stated | 2026-07-03 |
| Delores-Lin/MDPBench | Task ptMetric ptComparison conditions not established | 53 | MDPBench leaderboard Reported by a third party |
Evaluated revision not stated | 2026-07-03 |
| Delores-Lin/MDPBench | Task ruMetric ruComparison conditions not established | 36.2 | MDPBench leaderboard Reported by a third party |
Evaluated revision not stated | 2026-07-03 |
| Delores-Lin/MDPBench | Task thMetric thComparison conditions not established | 49.4 | MDPBench leaderboard Reported by a third party |
Evaluated revision not stated | 2026-07-03 |
| Delores-Lin/MDPBench | Task viMetric viComparison conditions not established | 54.1 | MDPBench leaderboard Reported by a third party |
Evaluated revision not stated | 2026-07-03 |
| Delores-Lin/MDPBench | Task zhMetric zhComparison conditions not established | 59.7 | MDPBench leaderboard Reported by a third party |
Evaluated revision not stated | 2026-07-03 |
| Delores-Lin/MDPBench | Task zh_tMetric zh_tComparison conditions not established | 59.2 | MDPBench leaderboard Reported by a third party |
Evaluated revision not stated | 2026-07-03 |
| PaddlePaddle/Real5-OmniDocBench | Task illuminationMetric illuminationComparison conditions not established | 78.1 | Real5-OmniDocBench Leaderboard Reported by a third party |
Evaluated revision not stated | 2026-08-08 |
| PaddlePaddle/Real5-OmniDocBench | Task overallMetric overallComparison conditions not established | 73.99 | Real5-OmniDocBench Leaderboard Reported by a third party |
Evaluated revision not stated | 2026-08-08 |
| PaddlePaddle/Real5-OmniDocBench | Task scanningMetric scanningComparison conditions not established | 86.17 | Real5-OmniDocBench Leaderboard Reported by a third party |
Evaluated revision not stated | 2026-08-08 |
| PaddlePaddle/Real5-OmniDocBench | Task screen_photographyMetric screen_photographyComparison conditions not established | 75.31 | Real5-OmniDocBench Leaderboard Reported by a third party |
Evaluated revision not stated | 2026-08-08 |
| PaddlePaddle/Real5-OmniDocBench | Task skewMetric skewComparison conditions not established | 63.01 | Real5-OmniDocBench Leaderboard Reported by a third party |
Evaluated revision not stated | 2026-08-08 |
| PaddlePaddle/Real5-OmniDocBench | Task warpingMetric warpingComparison conditions not established | 67.2 | Real5-OmniDocBench Leaderboard Reported by a third party |
Evaluated revision not stated | 2026-08-08 |
| allenai/olmOCR-bench | Task arxiv_mathMetric arxiv_mathComparison conditions not established | 77.2 | OlmOCR-Bench Github repository Reported by a third party |
Evaluated revision not stated | 2026-02-19 |
| allenai/olmOCR-bench | Task headers_footersMetric headers_footersComparison conditions not established | 96.1 | OlmOCR-Bench Github repository Reported by a third party |
Evaluated revision not stated | 2026-02-19 |
| allenai/olmOCR-bench | Task long_tiny_textMetric long_tiny_textComparison conditions not established | 79.4 | OlmOCR-Bench Github repository Reported by a third party |
Evaluated revision not stated | 2026-02-19 |
| allenai/olmOCR-bench | Task multi_columnMetric multi_columnComparison conditions not established | 66.4 | OlmOCR-Bench Github repository Reported by a third party |
Evaluated revision not stated | 2026-02-19 |
| allenai/olmOCR-bench | Task old_scansMetric old_scansComparison conditions not established | 33.3 | OlmOCR-Bench Github repository Reported by a third party |
Evaluated revision not stated | 2026-02-19 |
| allenai/olmOCR-bench | Task old_scans_mathMetric old_scans_mathComparison conditions not established | 73.6 | OlmOCR-Bench Github repository Reported by a third party |
Evaluated revision not stated | 2026-02-19 |
| allenai/olmOCR-bench | Task overallMetric overallComparison conditions not established | 75.7 | OlmOCR-Bench Github repository Reported by a third party |
Evaluated revision not stated | 2026-02-19 |
| allenai/olmOCR-bench | Task table_testsMetric table_testsComparison conditions not established | 80.2 | OlmOCR-Bench Github repository Reported by a third party |
Evaluated revision not stated | 2026-02-19 |
dots.ocr
| Benchmark | Conditions | Result | Reported by | Revision | Date |
|---|---|---|---|---|---|
| Delores-Lin/MDPBench | Task arMetric arComparison conditions not established | 75.9 | MDPBench leaderboard Reported by a third party |
Evaluated revision not stated | 2026-07-03 |
| Delores-Lin/MDPBench | Task deMetric deComparison conditions not established | 79.7 | MDPBench leaderboard Reported by a third party |
Evaluated revision not stated | 2026-07-03 |
| Delores-Lin/MDPBench | Task digitalMetric digitalComparison conditions not established | 88.8 | MDPBench leaderboard Reported by a third party |
Evaluated revision not stated | 2026-07-03 |
| Delores-Lin/MDPBench | Task enMetric enComparison conditions not established | 81.2 | MDPBench leaderboard Reported by a third party |
Evaluated revision not stated | 2026-07-03 |
| Delores-Lin/MDPBench | Task esMetric esComparison conditions not established | 69.2 | MDPBench leaderboard Reported by a third party |
Evaluated revision not stated | 2026-07-03 |
| Delores-Lin/MDPBench | Task frMetric frComparison conditions not established | 67.1 | MDPBench leaderboard Reported by a third party |
Evaluated revision not stated | 2026-07-03 |
| Delores-Lin/MDPBench | Task hiMetric hiComparison conditions not established | 77.3 | MDPBench leaderboard Reported by a third party |
Evaluated revision not stated | 2026-07-03 |
| Delores-Lin/MDPBench | Task idMetric idComparison conditions not established | 82.5 | MDPBench leaderboard Reported by a third party |
Evaluated revision not stated | 2026-07-03 |
| Delores-Lin/MDPBench | Task itMetric itComparison conditions not established | 87.8 | MDPBench leaderboard Reported by a third party |
Evaluated revision not stated | 2026-07-03 |
| Delores-Lin/MDPBench | Task jpMetric jpComparison conditions not established | 70.6 | MDPBench leaderboard Reported by a third party |
Evaluated revision not stated | 2026-07-03 |
| Delores-Lin/MDPBench | Task koMetric koComparison conditions not established | 68.5 | MDPBench leaderboard Reported by a third party |
Evaluated revision not stated | 2026-07-03 |
| Delores-Lin/MDPBench | Task latinMetric latinComparison conditions not established | 79.1 | MDPBench leaderboard Reported by a third party |
Evaluated revision not stated | 2026-07-03 |
| Delores-Lin/MDPBench | Task nlMetric nlComparison conditions not established | 78.8 | MDPBench leaderboard Reported by a third party |
Evaluated revision not stated | 2026-07-03 |
| Delores-Lin/MDPBench | Task non_latinMetric non_latinComparison conditions not established | 73.5 | MDPBench leaderboard Reported by a third party |
Evaluated revision not stated | 2026-07-03 |
| Delores-Lin/MDPBench | Task overallMetric overallComparison conditions not established | 76.5 | MDPBench leaderboard Reported by a third party |
Evaluated revision not stated | 2026-07-03 |
| Delores-Lin/MDPBench | Task photographedMetric photographedComparison conditions not established | 72.3 | MDPBench leaderboard Reported by a third party |
Evaluated revision not stated | 2026-07-03 |
| Delores-Lin/MDPBench | Task ptMetric ptComparison conditions not established | 86.9 | MDPBench leaderboard Reported by a third party |
Evaluated revision not stated | 2026-07-03 |
| Delores-Lin/MDPBench | Task ruMetric ruComparison conditions not established | 66.8 | Not named Reported by a third party |
Evaluated revision not stated | 2026-07-03 |
| PaddlePaddle/Real5-OmniDocBench | Task illuminationMetric illuminationComparison conditions not established | 87.57 | Real5-OmniDocBench Leaderboard Reported by a third party |
Evaluated revision not stated | 2026-08-08 |
| PaddlePaddle/Real5-OmniDocBench | Task overallMetric overallComparison conditions not established | 86.38 | Real5-OmniDocBench Leaderboard Reported by a third party |
Evaluated revision not stated | 2026-08-08 |
| PaddlePaddle/Real5-OmniDocBench | Task scanningMetric scanningComparison conditions not established | 86.87 | Real5-OmniDocBench Leaderboard Reported by a third party |
Evaluated revision not stated | 2026-08-08 |
| PaddlePaddle/Real5-OmniDocBench | Task screen_photographyMetric screen_photographyComparison conditions not established | 87.18 | Real5-OmniDocBench Leaderboard Reported by a third party |
Evaluated revision not stated | 2026-08-08 |
| PaddlePaddle/Real5-OmniDocBench | Task skewMetric skewComparison conditions not established | 84.27 | Real5-OmniDocBench Leaderboard Reported by a third party |
Evaluated revision not stated | 2026-08-08 |
| PaddlePaddle/Real5-OmniDocBench | Task warpingMetric warpingComparison conditions not established | 86.01 | Real5-OmniDocBench Leaderboard Reported by a third party |
Evaluated revision not stated | 2026-08-08 |
| allenai/olmOCR-bench | Task arxiv_mathMetric arxiv_mathComparison conditions not established | 82.1 | dots.ocr technical report Reported by a third party |
Evaluated revision not stated | 2026-02-19 |
| allenai/olmOCR-bench | Task headers_footersMetric headers_footersComparison conditions not established | 94.1 | dots.ocr technical report Reported by a third party |
Evaluated revision not stated | 2026-02-19 |
| allenai/olmOCR-bench | Task long_tiny_textMetric long_tiny_textComparison conditions not established | 81.2 | dots.ocr technical report Reported by a third party |
Evaluated revision not stated | 2026-02-19 |
| allenai/olmOCR-bench | Task multi_columnMetric multi_columnComparison conditions not established | 82.4 | dots.ocr technical report Reported by a third party |
Evaluated revision not stated | 2026-02-19 |
| allenai/olmOCR-bench | Task old_scansMetric old_scansComparison conditions not established | 40.9 | dots.ocr technical report Reported by a third party |
Evaluated revision not stated | 2026-02-19 |
| allenai/olmOCR-bench | Task old_scans_mathMetric old_scans_mathComparison conditions not established | 64.2 | dots.ocr technical report Reported by a third party |
Evaluated revision not stated | 2026-02-19 |
| allenai/olmOCR-bench | Task overallMetric overallComparison conditions not established | 79.1 | dots.ocr technical report Reported by a third party |
Evaluated revision not stated | 2026-02-19 |
| allenai/olmOCR-bench | Task table_testsMetric table_testsComparison conditions not established | 88.3 | dots.ocr technical report Reported by a third party |
Evaluated revision not stated | 2026-02-19 |
SAVRN's Notes on DeepSeek-OCR
Scanned pages, PDFs, any text trapped in an image: that is the job. Image plus prompt in, text out, from 3.3B parameters arranged as 64 routed experts with 6 active per token. At 16-bit the weights are 6.7 GB and it needs 8.0 GB, so the cheapest setup, one 192 GB MI300X at $1.85 an hour, sits mostly idle; batch documents hard or share the card with the language model that reads the output.
MIT allows commercial use, modification and redistribution provided the notices ride along. Check the runtime first: the publisher's tested path is the transformers library on NVIDIA GPUs with Python 3.12.9 and CUDA 11.8, with upstream vLLM support since October 23, 2025, so the MI300X the Index prices is a port you prove yourself. The 8,192 token context caps how much of a long document fits in one pass. The paper is arXiv:2510.18234.
SAVRN's Notes on dots.ocr
Layout detection and content recognition normally take two models. Dots Studio put both in one vision-language model: a 3B-parameter parser that reads a page, finds the blocks, pulls text and tables, and keeps the reading order, in several languages. It needs 7.3 GB at 16-bit, 3.6 GB at 8-bit, 1.8 GB at 4-bit. The cheapest listed setup is one MI300X with 192 GB at $1.85 per hour, so treat it as a sidecar beside the larger models in a document pipeline.
The MIT license is short: commercial use, modification and redistribution, with the copyright and permission notices kept. Three checks. The library field reads dots_ocr, its own code path, so confirm your serving stack loads it. Context is 131,072 tokens. The third-party MDPBench scores run from 88.8 on digital documents down to 67.1 for French, so test your own languages and scan quality before trusting the top line.
Questions
Which is larger, DeepSeek-OCR or dots.ocr?
DeepSeek-OCR (3.3B parameters) is larger than dots.ocr (3B parameters), by the parameter counts their publishers report.
Which is cheaper to run, DeepSeek-OCR or dots.ocr?
At 4-bit, DeepSeek-OCR fits on 1x MI300X from $1.85 an hour and dots.ocr on 1x MI300X from $1.85 an hour, at the lowest on-demand prices the SAVRN Index lists.
Can I use DeepSeek-OCR commercially?
Yes. DeepSeek-OCR is released under MIT License. The MIT License is a short permissive license. It permits commercial use, modification and redistribution, provided the copyright notice and permission notice are included.
Can I use dots.ocr commercially?
Yes. dots.ocr is released under MIT License. The MIT License is a short permissive license. It permits commercial use, modification and redistribution, provided the copyright notice and permission notice are included.