PQWERAI

Whitepaper

The technical whitepaper.

How PQWERAI works, what has been built and tested, what is still open, and what it takes to reach mainnet. Every claim points to its evidence.

Whitepaper: overview

Version 0.4.0

Status
TECHNICAL REVIEW DRAFT — author review, not an independent audit
Date
19 November 2026
Length
86 pages, PDF, 0.6 MiB
Source cutoff
repository 1bf413c; every claim links to its evidence in Appendix A
SHA-256
bb6ac3bb27d21ebee71cad0e8b9467e57617bba89213a579a0dfa1f89bea3fb8
Download
Download the whitepaper (PDF)   Open in a new tab

How to read it

The paper separates what is implemented, what was tested on one machine, what is deployed on the test network and what is only planned. Status words match the feature status page. Test PQAI and QAI have no monetary value. The whitepaper is a technical document: it is not an offer of tokens or any financial product, and it promises no mainnet date, allocation or return.

LaTeX source: docs/whitepaper/ in the project repository, with its evidence ledger and freshness checks.

What is inside

  • Architecture — GhostRider proof of work, Smartnode checkpoint finality, EVM and WASM, PASE parallel execution.
  • AI services — MeshLLM and Colibri inference outside consensus, escrowed AI jobs, model pools.
  • Economics — the 67/18/1/9/5 issuance split, the IPFS storage reward, QAI and the Quantum Reserve, with exact integer arithmetic.
  • Security and performance — threat model, open findings, measured throughput and what it does not show.
  • Path to mainnet — evidence-gated milestones, network growth targets and the resources the project needs.

New in 0.4.0

  • IPFS reward v2: bonded, slashable proof of access (switch at block 68,961).
  • PATE adaptive trust execution; phase 3 fast finality runs in shadow mode only.
  • AI market on testnet 20777, same-model compute pools, wallet mining on every CPU thread, advisory internal agents.
  • Throughput at the live difficulty: 15M, 30M and 60M gas, and batched payments.
  • A new chapter on the path to mainnet and how the project plans to resource it.