DeepSeek-OCR — local AI model by DeepSeek
Written by Jakub Rusinowski · Last updated
A 3B vision-language model built for one job — turning document pages into structured text — and a demonstration of "contexts optical compression": rendering text as an image and decoding it costs roughly 10x fewer tokens at near-lossless fidelity, or 20x at about 60% precision. A DeepEncoder compresses each page into vision tokens which a DeepSeek-3B MoE decoder reads, activating only ~570M parameters. A second generation, DeepSeek-OCR-2, followed in January 2026.
Variants
The smallest DeepSeek-OCR variant needs about 3 GB of VRAM at Q4_K_M — quantized weights plus framework overhead, before any KV cache.
| Model | VRAM at Q4 | VRAM | Context | Run it |
|---|---|---|---|---|
| DeepSeek-OCR 3B → 3B (0.57B active) | ~2.6 GB | 8,192 | ollama run deepseek-ocr |
Memory is quantized weights plus overhead at Q4_K_M, from the same engine as the GPU & VRAM checker.
How to run DeepSeek-OCR locally
Install Ollama, then pull the tag.
ollama run deepseek-ocrPick a size above for its own VRAM figure, speed estimate and install command.
Licence
Commercial use permitted. No usage restrictions beyond attribution.
Applies to: DeepSeek-OCR 3BRecommended GPU
The cheapest catalogued GPU that runs DeepSeek-OCR locally (min 3 GB VRAM) is the Intel Arc B570 (10 GB).
Can I run DeepSeek-OCR on my GPU?
DeepSeek-OCR — frequently asked questions
How much VRAM does DeepSeek-OCR need?
DeepSeek-OCR needs about 3 GB VRAM at Q4_K_M quantization for its smallest variant. Variants: DeepSeek-OCR 3B (3 GB, Q4_K_M). On Apple Silicon, unified memory counts toward this requirement.
Can I run DeepSeek-OCR on an RTX 4090 (24 GB)?
Yes — DeepSeek-OCR runs on an RTX 4090 (24 GB) and other 24 GB cards such as the RTX 3090. Smaller variants also fit comfortably on 8–16 GB GPUs at Q4_K_M.
What quantization should I use for DeepSeek-OCR?
Q4_K_M is the best balance of quality and VRAM for DeepSeek-OCR in most cases. Choose Q8_0 for near-lossless quality if you have spare VRAM, or smaller quants (Q3/Q2) only when memory is tight.
How do I run DeepSeek-OCR with Ollama?
DeepSeek-OCR has no local Ollama tag — the published tag is cloud-hosted, so running it sends your prompts to a hosted GPU rather than your own machine.