Published by deepseek-ai. 1 member measured on real machines, across 1 build. Every number below was signed by the machine that produced it — none of it is from a model card.
what it takes, what it returns
image + text → text
Declared by the network’s catalog, not measured — a signature says what this family would be asked. Whether it answered is the signed capability flags on each member.
13.3 → 340.2 tok/s across 2 machines · ×25.5 spreadunnamed machinex86 · CPU
Builds this network knows of
Declared, not measured — this is what the catalog says exists and what it would cost a machine to hold. Nothing here is signed, and none of it claims the build works.
how much memory does your machine have
29 of 29 builds fit a 16 GB machine — budgeting 70% of it, or 11.2 GiB, because the cache and the runtime want the rest.
hf.co/NexaAI/DeepSeek-OCR-GGUF:Q4_0Q4_0ollamaquantized by NexaAI
1.5 GiB
hf.co/NexaAI/DeepSeek-OCR-GGUF-CUDA:Q4_0Q4_0ollamaderived by NexaAI
1.5 GiB
hf.co/sabafallah/DeepSeek-OCR-GGUF:Q4_K_MQ4_K_Mollamameasured herequantized by sabafallah
1.8 GiB
hf.co/NexaAI/DeepSeek-OCR-GGUF:Q5_0Q5_0ollamaquantized by NexaAI
1.9 GiB
hf.co/NexaAI/DeepSeek-OCR-GGUF-CUDA:Q5_0Q5_0ollamaderived by NexaAI
1.9 GiB
mlx-community/DeepSeek-OCR-4bit4BITmlxderived by mlx-community
2.3 GiB
aoiandroid/mirror-mlx-community-DeepSeek-OCR-4bit4BITmlxderived by aoiandroid
2.3 GiB
hf.co/NexaAI/DeepSeek-OCR-GGUF-CUDA:Q8_0Q8_0ollamaderived by NexaAI
2.9 GiB
hf.co/sabafallah/DeepSeek-OCR-GGUF:Q8_0Q8_0ollamaquantized by sabafallah
2.9 GiB
hf.co/ggml-org/DeepSeek-OCR-GGUF:Q8_0Q8_0ollamaderived by ggml-org
2.9 GiB
mlx-community/DeepSeek-OCR-6bit6BITmlxderived by mlx-community
3.0 GiB
mlx-community/DeepSeek-OCR-8bit8BITmlxderived by mlx-community
3.6 GiB
hf.co/mohammad2928git/DeepSeek-OCR-GGUF:BF16BF16ollamaderived by mohammad2928git
5.5 GiB
hf.co/NexaAI/DeepSeek-OCR-GGUF-CUDA:BF16BF16ollamaderived by NexaAI
5.5 GiB
hf.co/NexaAI/DeepSeek-OCR-GGUF-CUDA:F16F16ollamaderived by NexaAI
5.5 GiB
hf.co/sabafallah/DeepSeek-OCR-GGUF:BF16BF16ollamaquantized by sabafallah
5.5 GiB
deepseek-ai/DeepSeek-OCRBF16vllmpublished by deepseek-ai
6.2 GiB
amd/DeepSeek-OCRBF16vllmderived by amd
6.2 GiB
colaiby2/DeepSeek-OCRBF16vllmderived by colaiby2
6.2 GiB
Jinstudio/DeepSeek-OCRBF16vllmderived by Jinstudio
6.2 GiB
kp-forks/DeepSeek-OCRBF16vllmderived by kp-forks
6.2 GiB
lvyufeng/DeepSeek-OCRBF16vllmderived by lvyufeng
6.2 GiB
OpenKing/DeepSeek-OCRBF16vllmderived by OpenKing
6.2 GiB
prince-canuma/DeepSeek-OCRBF16vllmderived by prince-canuma
6.2 GiB
Sci-fi-vy/DeepSeek-OCRBF16vllmderived by Sci-fi-vy
6.2 GiB
tphage/DeepSeek-OCRBF16vllmderived by tphage
6.2 GiB
unsloth/DeepSeek-OCRBF16vllmderived by unsloth
6.2 GiB
whrrr/DeepSeek-OCRBF16vllmderived by whrrr
6.2 GiB
not-lain/finetuned_deepseek_ocrBF16vllmderived by not-lain
6.2 GiB
Sizes are what the catalog declares, not what a machine measured — fitting on disk is necessary and not sufficient. Nobody here has run these on your machine, and this narrows the candidates rather than certifying one.
the same builds, by member
DeepSeek-OCR BASE
hf.co/NexaAI/DeepSeek-OCR-GGUF:Q4_0Q4_0ollama1.5 GiB on diskquantized by NexaAI
hf.co/sabafallah/DeepSeek-OCR-GGUF:Q4_K_MQ4_K_Mollama1.8 GiB on diskquantized by sabafallah
hf.co/NexaAI/DeepSeek-OCR-GGUF:Q5_0Q5_0ollama1.9 GiB on diskquantized by NexaAI
hf.co/sabafallah/DeepSeek-OCR-GGUF:Q8_0Q8_0ollama2.9 GiB on diskquantized by sabafallah
hf.co/sabafallah/DeepSeek-OCR-GGUF:BF16BF16ollama5.5 GiB on diskquantized by sabafallah
deepseek-ai/DeepSeek-OCRBF16vllm6.2 GiB on diskpublished by deepseek-ai
DeepSeek-OCR
amd/DeepSeek-OCRBF16vllm6.2 GiB on diskderived by amd
DeepSeek-OCR
colaiby2/DeepSeek-OCRBF16vllm6.2 GiB on diskderived by colaiby2
DeepSeek-OCR-GGUF
hf.co/ggml-org/DeepSeek-OCR-GGUF:Q8_0Q8_0ollama2.9 GiB on diskderived by ggml-org
DeepSeek-OCR
Jinstudio/DeepSeek-OCRBF16vllm6.2 GiB on diskderived by Jinstudio
DeepSeek-OCR
kp-forks/DeepSeek-OCRBF16vllm6.2 GiB on diskderived by kp-forks
DeepSeek-OCR
lvyufeng/DeepSeek-OCRBF16vllm6.2 GiB on diskderived by lvyufeng
DeepSeek-OCR-8bit
mlx-community/DeepSeek-OCR-4bit4BITmlx2.3 GiB on diskderived by mlx-community
mlx-community/DeepSeek-OCR-6bit6BITmlx3.0 GiB on diskderived by mlx-community
mlx-community/DeepSeek-OCR-8bit8BITmlx3.6 GiB on diskderived by mlx-community
DeepSeek-OCR-GGUF_v1
hf.co/mohammad2928git/DeepSeek-OCR-GGUF:BF16BF16ollama5.5 GiB on diskderived by mohammad2928git
DeepSeek-OCR-GGUF-CUDA
hf.co/NexaAI/DeepSeek-OCR-GGUF-CUDA:Q4_0Q4_0ollama1.5 GiB on diskderived by NexaAI
hf.co/NexaAI/DeepSeek-OCR-GGUF-CUDA:Q5_0Q5_0ollama1.9 GiB on diskderived by NexaAI
hf.co/NexaAI/DeepSeek-OCR-GGUF-CUDA:Q8_0Q8_0ollama2.9 GiB on diskderived by NexaAI
hf.co/NexaAI/DeepSeek-OCR-GGUF-CUDA:BF16BF16ollama5.5 GiB on diskderived by NexaAI
hf.co/NexaAI/DeepSeek-OCR-GGUF-CUDA:F16F16ollama5.5 GiB on diskderived by NexaAI
finetuned_deepseek_ocr
not-lain/finetuned_deepseek_ocrBF16vllm6.2 GiB on diskderived by not-lain
DeepSeek-OCR
OpenKing/DeepSeek-OCRBF16vllm6.2 GiB on diskderived by OpenKing
DeepSeek-OCR
prince-canuma/DeepSeek-OCRBF16vllm6.2 GiB on diskderived by prince-canuma
DeepSeek-OCR
Sci-fi-vy/DeepSeek-OCRBF16vllm6.2 GiB on diskderived by Sci-fi-vy
DeepSeek-OCR
tphage/DeepSeek-OCRBF16vllm6.2 GiB on diskderived by tphage
DeepSeek-OCR
unsloth/DeepSeek-OCRBF16vllm6.2 GiB on diskderived by unsloth
DeepSeek-OCR
whrrr/DeepSeek-OCRBF16vllm6.2 GiB on diskderived by whrrr
mirror-mlx-community-DeepSeek-OCR-4bit
aoiandroid/mirror-mlx-community-DeepSeek-OCR-4bit4BITmlx2.3 GiB on diskderived by aoiandroid
ask your own gram
Your concierge can answer this from the network rather than from this page. The ask carries the operator id, so it resolves to a capability and an engine — not to a name somebody has to recognise.
What has this network measured about the DeepSeek-OCR family (epn.family.deepseek-ocr), and can my gram run any of it? See /inferences/deepseek-ocr
