Skip to main content

The physics virtual machine

Zeq is a physics virtual machine. Real physical laws are the instruction set — you compute against 1,600+ operators across 65 domains, from quantum to classical to relativistic, on one computational surface with no translation layer between regimes. Every computation is synchronized to the 1.287 Hz HulyaPulse — one tick is one Zeqond (0.777 s) — and resolved to ≤0.1% precision. And because the whole surface is deterministic and clock-locked, a result comes back byte-identical across every node and carries a proof any node re-verifies: physics you can reproduce and check, not a floating-point number you have to trust.

It's physics infrastructure, not an AI product. AI is one consumer of it — through the harness, an MCP surface, and a kernel skill — but the substrate underneath is synchronized, verifiable physics.

Physics-native, on one clock

One surface, every regime

Quantum, classical, and relativistic physics on a single computational surface — 1,600+ operators, no translation layer between regimes.

Synchronized to 1.287 Hz

Every computation runs on the HulyaPulse — one tick is one Zeqond (0.777 s) — so all of physics shares a single, exact time base.

Deterministic & provable

Byte-identical across every node, resolved to ≤0.1%, with a ZeqProof on every result — reproducible physics, not floating-point drift you must trust.

Compute something real, right now

No signup. Mint a free key and derive the Schwarzschild radius of the Sun — from r_s = 2GM/c², not a lookup:

ZEQ_KEY=$(curl -sX POST https://zeqsdk.com/api/demo-key/mint | jq -r .key)

curl -sX POST https://zeqsdk.com/api/zeq/compute \
-H "Authorization: Bearer $ZEQ_KEY" -H "Content-Type: application/json" \
-d '{"operators":["KO42","GR37"],"inputs":{"mass":1.98892e30}}'
# → 2954.0077 m — with the equation, the bound constants, an uncertainty, and a signed proof

Swap the operator and you have a different science: MED_BMI {mass:70,height:1.75}22.857 kg/m² · BLACK_SCHOLES {S:100,K:100,r:0.05,T:1,sigma:0.2}10.4506 · KEPLER_THIRD {a:1.496e11,M:1.989e30}≈ 1 year. Watch the clock itself live: GET https://zeqsdk.com/api/zeq/pulse (public).

Want to see exactly how that number is produced on the CPU — no black box? Follow one real request step by step in the execution model.

Three primitives

Operators — the vocabulary

Every compute names the operators it runs, every contract names the operator each transition fires, every audit row names the operator it proves. The catalogue is 1,600+ operators across 65 categories — quantum, relativity, fluid dynamics, condensed matter, finance, signal processing, and more — each a named, standard formula the engine evaluates for real:

OperatorIsA call returns
NM19F = ma{m:10,a:3}30 N
GR37r_s = 2GM/c²{mass:1.99e30}2954 m
QM9de Broglie λ = h/p{m:9.109e-31,v:1e6}7.27×10⁻¹⁰ m
CMP4conductivity σ = neμ{n:1e23,mu:0.1}1602 S/m

An unknown operator ID is refused (UNKNOWN_OPERATOR), never faked — the registry is the type. Browse the whole catalogue in the operator reference, each category labelled with how it computes (dedicated closed form vs the transparent ODE fallback).

State contracts — logic you can execute

A state contract turns operators into running logic: named states, transitions that fire operators under conditions, and a ZeqProof per proof-required fire — all on the machine's audit chain. 316 templates ship ready-made across 45 categories. Deploy one, fire it, watch it:

# deploy a force-threshold watchdog (NM19 on every reading) into your machine
"name":"zeq_contract_deploy", "arguments":{"templateId":"force-threshold-alarm"}
# fire a transition — it lands on your audit chain, clock-stamped
"name":"zeq_contract_fire", "arguments":{"id":"CONTRACT_ID","input":{}}

The full walkthrough — verified end to end against a live node — is in Deploy your first contract; the mechanics are in anatomy and the lifecycle.

Why you can trust a result

Zeq's doctrine is never say "trust me." Every compute returns a signed envelope carrying a ZeqProof (HMAC(operator | result | zeqond)); any node can re-verify it. Private machine histories are made tamper-resistant by a Proof-of-Elapsed-Zeqonds seal spine — rewriting the past costs real sequential time. And the honesty contract means the machine computes, asks for what it needs, or refuses — it never fabricates.

Applications built on Zeq

Real apps, running on machines, computing through the same surface these docs describe:

More, all live: ZeqSSL, ZSP, BYOK, Audit Daemon, Skills, TESC, Message, Edge. See the full applications gallery.

Explore the docs


Everything on this site is generated from or verified against the live system — the endpoint reference from the node's own route registry, the operators from the live registry, and every worked example prints a number you can reproduce with the exact command shown. If a page fails that bar, it's a bug.