Mechanism

The Neural VM.

Ternary models run in chain state, in integers, byte for byte.

Model ClassBitNet-class ternary
ArithmeticPure integer
DecodingGreedy · committed seed
GuaranteeByte-identical
Ceiling~2-3B, under 1GB
01Ternary Weights

Add, subtract or skip

01 · Weights of −1, 0, +1

Three values per weight

In a BitNet-class model every weight is −1, 0 or +1. A matrix multiply stops being floating-point arithmetic and becomes additions, subtractions and skips over integer activations. Integers add the same way on every chip, in every order, so the result is exact.

  • +1 adds the activation to the running sum
  • −1 subtracts it
  • 0 skips it: no work, no rounding
02Byte-Identical by Construction

Integers don't drift

The same model on three unlike machines. In floating point, low bits differ by chip and batch; in pure integer, the hashes are one.

02 · The AIOS Module

A pinned runtime

Every Miner runs the AIOS Module: pure-integer fixed-point kernels that give the same result whatever the batch size, with greedy or committed-seed decoding. It is pinned by content hash, so a node running anything else simply does not match. Cross-GPU floating-point byte identity is an open research problem; AIOS does not depend on it.

03The Ceiling, Stated

Small, and useful

What limits a model in chain state is state size and per-block compute, not model quality.

03 · Models Live on Blocks

About 2 to 3 billion parameters

A natively trained ternary model of that size fits in under 1GB and is benchmark-competitive at its size. That is the credible ceiling for a model living in chain state. Larger models run off-chain across the Miner network and verify statistically.

  1. 01 / Load

    Weights from chain state

    The model is content-addressed state. Every node holds the same bytes.

  2. 02 / Execute

    Integer kernels, pinned

    The Neural VM runs the layers in the AIOS Module, deterministic by construction.

  3. 03 / Decode

    Greedy or a committed seed

    Sampling randomness is committed up front, so the output is reproducible.

  4. 04 / Commit

    One output hash

    Every conforming node lands on the same hash, and the committee can check it.

Neural VM · Spec
Model class
BitNet-class, 1-bit and ternary, integer-quantized
Arithmetic
Pure integer, batch-invariant kernels, no IEEE-754
Decoding
Greedy or committed seed, deterministic by construction
Guarantee
Byte-identical output on every conforming node
Ceiling
About 2 to 3 billion parameters natively ternary, under 1GB
Never
A frontier chatbot in consensus. Floating-point models verify statistically.
Weights Are Programs

Same input, same bytes

That one property is what turns agreement into something a chain can check.