21 NAND Docs 中文

Gate brain

The small model BitCPM4-1B, turned entirely into NAND gates. Your computer's GPU computes the answer gate by gate, bit for bit identical to the model's integer reference implementation.

What you need

If your device isn't enough, the page checks first and tells you. It doesn't download a single byte.

How to use it

  1. Open the page and click "Get ready".
  2. Wait while it downloads the model and builds the gate circuit on your computer. This usually takes 2–8 minutes; newer computers are faster. Don't close the page while it gets ready.
  3. When it's ready, type a question at the bottom and click "Ask".
  4. Next time you open it, it's ready in a dozen seconds or so, even offline. The built circuit is kept in the browser's local storage.

Your questions are not sent to any server. Everything is computed on your computer.

Some numbers

ItemValue
ModelBitCPM4-1B (OpenBMB, Apache-2.0)
Gates computed per wordabout 258.3 billion
"Memory switches" that hold the contextabout 109 million
Question plus answer, at most128 tokens
Long-answer modeabout 512 tokens; needs a new setup, about 7.6 GB of memory, and each word is slower
First downloadabout 580 MB

The page converts between text and token numbers itself. That part is not done in gates.

How you know it computes correctly

Follow-up questions: tested, and it often loses the thread

The page sends the earlier questions and answers along with the next question, but the model often forgets what was said earlier. What we tested and what came out (2026-11-10):

What was testedResultLine to passVerdict
20 two-turn items where the second question cannot be answered without the first turn (default mode)12 / 20 right14Fail
The same second questions asked alone, without the earlier turn (control)1 / 20 right by luckat most 4Pass
10 four-turn dialogues, 3 questions each that need the earlier turns (long-answer mode)23 of 30 right; 8 dialogues with at least 2 rightall 10 with at least 2 rightFail
Those 30 questions asked alone (control)4 / 30 right by luckat most 6Pass
The same inputs run again in a fresh process40 / 40 answers identical token for token40 / 40Pass

Progress

DateWhat happened
2026-09-24The whole model turned into gates, answering on its own in the web page, bit for bit identical to the model's integer reference implementation
2026-09-26Gates per word cut from about 1.4 trillion to about 260 billion, results still bit for bit the same
The whole model is not put on chain: it is ten million times larger than one on-chain circuit can hold. For the cost of running a large model on chain, see What a large model on chain would cost.

Next