family member
DeepSeek-OCR 2 logo

DeepSeek-OCR 2 BASE

412 tokens a second on Windows · GeForce RTX 3080 Ti 12 GB · 16 CPUs · 16 GB RAM. No hosted API sells it, so this network is the place to run it.

1 build of DeepSeek-OCR 2 BASE, 3 signed runs on real machines: different packagers, precisions and engines, each a different file with its own numbers.

412tok/s, fastest machine
$0.63per M tokens · no API sells it
5.8 GBmemory, smallest build
3providers measured it
Run it on your gram Install a gram with one command. It works through the catalog on your own machine and signs every result. 3 providers have measured DeepSeek-OCR 2 BASE so far.Use it in your own room Ask Evo in your gramx and it runs on a machine that proved it, at the metered $0.63 per million tokens. On your own machines it costs you nothing.Compare every build Put it beside every other model in one table: speed, price, context and the API price, one row each.
best on ollama · 1 build

no context proved on this engine yet

411.7 tok/s peak, by sabafallah Q4_K_M

Speed against price, every machine

Up and left is faster and cheaper · the line joins machines no other beats on both · price on a log scale

$1USD per million tokenstok/sWindows · GeForce RTX 3080 Ti 12 GB · 16 CPUs · 16 GB RAM · 4-bit by sabafallah · 411.73 tok/s · $0.626/MLinux · L4 23 GB · 12 CPUs · 53 GB RAM · 4-bit by sabafallah · 334.82 tok/s · $0.6487/MWindows · Intel(R) Iris(R) Xe Graphics 1 GB · 8 CPUs · 16 GB RAM · 4-bit by sabafallah · 15.36 tok/s · $4.7185/M

Same build, different machines

Q4_K_M by sabafallah on 3 machines: the fastest decodes 27× as fast as the slowest. One dot per machine, tokens per second.

  1. NVIDIA GeForce RTX 3080 Ti×1412 tok/s
  2. NVIDIA L4×1335 tok/s
  3. Intel(R) Iris(R) Xe Graphics×115 tok/s

What it fits on

The smallest build needs about 5.8 GB; 15 of 18 kinds of machine on the network can hold it. Bars: each kind's memory (its card's, else RAM) · ticks: every build, at the size it held plus a fifth for the prompt.

  1. Linux, CPU only · 2 CPUs1 GB RAMnone
  2. NVIDIA GeForce RTX 3050 Laptop GPU4 GB on the cardnone
  3. NVIDIA GeForce GTX 16504 GB on the cardnone
  4. AMD Radeon(TM) Vega 8 Mobile Graphics7 GB RAMevery build
  5. Windows, CPU only · 12 CPUs7 GB RAMevery build
  6. Apple silicon8 GB RAMevery build
  7. Apple M18 GB RAMevery build
  8. Mac, CPU only · 8 CPUs8 GB RAMevery build
  9. NVIDIA GeForce RTX 3080 Ti12 GB on the cardevery build
  10. Tesla T415 GB on the cardevery build
  11. Intel(R) UHD Graphics 73016 GB RAMevery build
  12. Intel(R) Iris(R) Xe Graphics16 GB RAMevery build
  13. Apple M416 GB RAMevery build
  14. Apple M116 GB RAMevery build
  15. NVIDIA L422 GB on the cardevery build
  16. 2× Tesla T429 GB on the cardevery build
  17. Linux, CPU only · 4 CPUs31 GB RAMevery build
  18. NVIDIA RTX PRO 6000 Blackwell Server Edition95 GB on the cardevery build

How far back it remembers, machine by machine

Filled: the needle came back at that prompt size · outlined: asked and lost · blank: not asked · trained for 8K. 3 walks ended because the model forgot, 0 because the machine could not hold the prompt.

machine · build4Kwhy it stopped
Gram ·84bf54NVIDIA GeForce RTX 3080 Ti · Q4_K_M by sabafallah— the model lost the needle at 4K
Gram ·dae679NVIDIA L4 · Q4_K_M by sabafallah— the model lost the needle at 4K
Gram ·2e5c02Intel(R) Iris(R) Xe Graphics · Q4_K_M by sabafallah— the model lost the needle at 4K

Carbon per million tokens

Lowest build on each machine · 2 of 3 read off a power sensor, the rest modelled from the parts · fewer grams is better

  1. Gram ·dae679314 g2.39 J/tok · sensor
  2. Gram ·84bf54781 g3.97 J/tok · sensor
  3. Gram ·2e5c023.0 kg15.41 J/tok · modelled

Carbon is location-based: signed energy × the yearly average of the grid where the machine signed its place (world grid average (Singapore not in the table) · India grid average; Ember Global Electricity Review 2025 (2024 generation), location-based). Facility overhead is not included. A machine on its own solar is not credited until it can prove it.

Measured by

The people whose machines signed these numbers

proved by at least one build: Structured output · Sustained output · Reasons about code

Every build

Each its own file with its own page: the machines that ran it and their signatures.

buildenginewhose workmemorycontext provedbest tok/smachines
Q4_K_M by sabafallahhf.co/sabafallah/DeepSeek-OCR-2-GGUF:Q4_K_Mollamaquantizer · sabafallah5.8 GB—411.73
derivatives — somebody else's change to these weights

Fine-tunes, merges and abliterations of this member, each its own model with its own page. None of their numbers are counted above.

DeepSeek-OCR-2-Q5_K_M.ggufnot measuredDeepSeek-OCR-2not measuredAX-DeepSeek-OCR-2-MLX-AXQ-6bitnot measuredAX-DeepSeek-OCR-2-CUDA-AXQ-NVFP4-W4A16not measuredDeepSeek-OCR-2not measuredUniversal-DeepSeek-OCR-2not measuredDeepSeek-OCR-2-Q4_K_M-GGUFnot measuredDeepSeek-OCR-2-MXFP8-CT-AutoRoundnot measuredArabic-OCR-DeepSeek-OCR-2not measuredDeepSeek-OCR-2-8bitnot measureddeepseek-ocr-2-transformers-v4.57.1not measuredDeepSeek-OCR-2not measuredDeepSeek-OCR-2not measured