Q
Quilibrium
Mainnet V2.1 Live

The Decentralized
Internet Layer

Secure every bit of web traffic through cryptographic means. Complete privacy, scalable computation, and zero-knowledge infrastructure for the next generation of the internet.

100M+
Messages/sec
0
% Fair Launch
Storage Node
Compute Node
Network Node
Validator
Network Active | 4 Regions

See It In Action

Watch the Quilibrium explanation video to learn about the next generation of private, decentralized computing.

Quilibrium Network Demo

Explore the architecture powering the decentralized internet layer

What is Quilibrium?

A simple explanation of the decentralized internet layer and how it works.

ELI5: Quilibrium Explained

Learn the basics in simple terms

Ready to Build the Future?

Join the decentralized internet layer that prioritizes privacy, scalability, and true cryptographic security.

Architectural Excellence

Built from the ground up for privacy, scalability, and true decentralization. Not a blockchainโ€”a complete reinvention.

Native Privacy

Zero-knowledge by design. The network cannot see computations or data. Complete nescience through MPC and encrypted storage.

Timechain Architecture

VDF-linked temporal ordering enables parallel execution and cryptographic proof sequencing without blocks or linear chains.

Hypergraph Data

RDF semantic validation supports complex relationships and knowledge graphs impossible on traditional blockchains.

RPM Mixnets

Random Permutation Matrix routing provides traffic analysis resistance beyond Tor, with complete sender-receiver unlinkability.

Horizontal Sharding

Shared-nothing architecture scales to petabyte storage and 200k+ messages/sec without compromising decentralization.

Proof of Meaningful Work

Anti-ASIC consensus rewards useful computation. Hardware diversity ensures broad participation from Raspberry Pi to servers.

Competitive Analysis

See how Quilibrium stacks up against major blockchain platforms across key dimensions.

Token Supply Comparison

Comparative analysis of supply models across major blockchain networks. Quilibrium features a dynamic, demand-responsive emission schedule.

Bitcoin

BTC
Max 21M
Circ ~19.8M
Model Fixed

Ethereum

ETH
Max โˆž
Circ ~120M
Model Deflation

Solana

SOL
Max โˆž
Circ ~621M
Model Disinflate

ICP

ICP
Max โˆž
Circ ~549M
Model Deflation

Bittensor

TAO
Max 21M
Circ ~11M
Model Halving (BTC-like)
Q

Quilibrium

QUIL
Max ~1.75B
Circ ~1.35B
Model Dynamic

Supply Model Comparison

BTC Fixed 21M cap
21,000,000
Current: ~19.8M 2140: 100%
ETH Unlimited with burn
โˆž Unlimited
Current: ~120M Net deflationary
SOL Disinflationary
โˆž Unlimited
Current: ~621M Target: ~1.5%
ICP Deflationary via cycles
โˆž Unlimited
Current: ~549M Burning reduces
TAO Bitcoin-like halving
21M Fixed Cap
Current: ~11M (51%) Halving every 4 years
QUIL Dynamic demand-responsive
~1.75B (2035)
Current: ~1.35B (77%) 2035: Convergence

Adaptive Emission

Unlike fixed schedules, Quilibrium adjusts issuance based on network difficulty and demand

Sustainable Security

Prevents "shrinking security budget" through work-proportional rewards

Fair Distribution

100% fair launch with no premine, VC allocation, or team reserves

Detailed Tokenomics Comparison

Feature Bitcoin Ethereum Solana ICP Bittensor Quilibrium
Supply Model Fixed Cap Unlimited + Burn Disinflationary Deflationary Fixed 21M Cap Dynamic/Adaptive
Current Supply ~19.9M ~120M ~621M ~549M ~11M ~1.35B
Max Supply 21M Unlimited Unlimited Unlimited 21M ~1.75B (2035)
Emission Schedule Halving (4yr) Proof of Stake Disinflationary Node rewards Yuma Consensus (PoI) Difficulty-triggered
Team/VC Allocation 0% (Fair) Premine + ICO ~15% Foundation Significant Founder allocation <1% (Fair)
Value Accrual Scarcity Burn + Staking Network usage Cycle burning AI model quality Computation demand
Consensus Proof of Meaningful Work
Cryptography BLS48-581 / Ed448
Throughput 1,000,000+ msg/sec
Finality Soft/Hard/Economic Tiers
Privacy Native MPC + Mixnets
Quantum Resistance Post-Quantum Ready

Start Mining

Join the network by running a Quilibrium node. Earn rewards through Proof of Meaningful Work while contributing to a decentralized internet layer.

System Requirements

CPU

Multi-core processor (ARM64 or x86_64)

More cores = more workers

Storage

SSD recommended for worker data

Regular backups required

Network

Stable internet connection

Specific ports required

Memory

Sufficient RAM for workers

Scales with worker count

Port Protocol Purpose
8336 QUIC/UDP or TCP Master process P2P communication
8340 TCP Master process streaming
25000-25003 QUIC/UDP or TCP Worker processes P2P (per worker)
32500-32503 TCP Worker processes streaming (per worker)

Important Security Notes

  • Backup your config.yml and keys.yml files securely - they are required to access rewards
  • Regularly backup worker data from worker-store/ to avoid penalties
  • Hosting providers may require firewall rules to block private IP ranges - see documentation

Token Economics

Dynamic, demand-responsive supply model with 100% fair launch distribution. Three emission scenarios model miner participation and network growth through 2033.

1.35B
Current Supply (2026)
Starting point
1.57B
Reduced Miners (2033)
Lower participation
1.59B
Expected Miners (2033)
Baseline projection
1.72B
Higher Miners (2033)
High participation

Emissions & Supply Curve

Projected QUIL supply and World State growth (2026โ€“2033)

Live Projection
Reduced Miners Supply
Expected Miners Supply
Higher Miners Supply
Expected World State (GB)

Reduced Miners

1.57B QUIL by 2033

Lower miner participation leads to slower emission growth. Supply rises from 1.35B to 1.57B, with the steepest climb between 2026โ€“2028 before plateauing as difficulty adjusts.

+220M

Expected Miners

1.59B QUIL by 2033

Baseline scenario with moderate participation. Supply reaches 1.59B by 2033. The curve follows natural network adoption with ASERTi3-2d difficulty adjustment ensuring sustainable incentives.

+240M

Higher Miners

1.72B QUIL by 2033

High participation accelerates emissions. Supply grows to 1.72B by 2033. The extended growth period reflects increased network difficulty and generational milestone unlocks driving faster token issuance.

+370M

World State Growth

143 GB โ†’ 1.8B GB by 2033

The World State represents total data stored across the Quilibrium network. Starting at 143 GB in early 2026, it grows linearly to 1.8 Billion GB by 2033 as the network scales to handle petabyte-level storage across horizontal shards.

~12,587x growth

How It Works

Unlike fixed-schedule cryptocurrencies, Quilibrium uses ASERTi3-2d difficulty adjustment with generational milestones. New emission phases unlock when network-wide difficulty reaches thresholds, ensuring sustainable incentives as technology advances.

  • No premine, no VC allocation, no airdrops
  • Work-proportional rewards prevent "shrinking security budget"
  • Supply curves steepen 2026โ€“2028, then plateau toward convergence

Scenario Breakdown

Reduced Miners
Lower participation โ†’ Slower rise to ~1.57B by 2033
Expected Miners
Baseline projection โ†’ Rise to ~1.59B by 2033
Higher Miners
High participation โ†’ Rise to ~1.72B by 2033
World State Growth
Expected data storage โ†’ ~1.8B GB by 2033

Shard Analytics

Real-time network sharding analysis. Shards are grouped by ring level based on worker count, indicating competition intensity and reward potential.

Global Shard Matrix

Interactive 3D visualization of Quilibrium network shards

Live Network

Active Workers Only: Shows only confirmed, participating workers. Excludes joining, leaving, and rejected states.

Global Distribution

Shards visualized across a 3D globe with real-time positioning

Interactive View

Rotate, zoom, and click shards to explore network details

Live Data

Real-time shard metrics from Quilibrium Explorer API

Ring 0 - Optimal
Ring 1 - Good
Ring 2 - Fair
Ring 3 - Poor
Ring 4+ - Overloaded
Unassigned

Version Distribution

Auto-detects latest version from network | Shows current distribution

Total Peers
--
Latest Release
--
highest build detected
Current Most Used
--
-- peers
Upgrade Progress
--%
peers on latest
Rank Version Build (Dec) Peers Percentage Status Distribution
Loading version distribution...

Version Analysis

  • Loading analysis...

Ring Level Overview

Shards grouped by competition intensity (Ring = Workers รท 8)

Shard Size by Ring

Understanding Rings

0
Ring 0
0-7 workers. Lowest competition, highest reward potential.
1
Ring 1
8-15 workers. Moderate competition, balanced rewards.
2
Ring 2
16-23 workers. High competition, diminishing returns.
3+
Ring 3+
24+ workers. Very high competition, lower rewards.
Unassigned
No workers. Opportunity if data size is significant.

Mining Strategy Insights

  • Loading analysis...
E2EE Machine Learning

Klearu

End-to-end encrypted ML primitives and runtime. Private inference, sub-linear training, and secure multi-party computation for the next generation of AI.

10
Workspace Crates
SLIDE
Sub-linear Training
2PC
Private Inference
Rust
Native Performance

Modular Architecture

10 specialized crates working in harmony for private, efficient machine learning

klearu-core

Foundation crate with LSH hash families, sparse tensors, and SLIDE network training primitives.

SimHash WtaHash MinHash

klearu-accel

Hardware acceleration with SIMD vectorization, BF16 quantization, and cache-aligned memory.

AVX2 NEON BF16

klearu-mongoose

Learnable hash functions with adaptive rebuild scheduling and drift detection.

Adaptive EMA

klearu-bolt

LSH hyperparameter autotuning and sparse inference optimizations.

Autotune Recall 90%

klearu-dejavu

Transformer sparsity prediction for attention heads and MLP neurons.

Sparsity Prediction

klearu-llm

LLaMA-compatible LLM inference engine with GQA, RoPE, RMSNorm, and SwiGLU.

LLaMA Inference

klearu-dpf

Distributed Point Functions using AES-based BGI construction and DCF.

DPF AES

klearu-mpc

Two-party computation with fixed-point arithmetic and Beaver triples.

2PC Q16.16

klearu-private

End-to-end private LLM inference via 2PC with Ferret OT and Ristretto255 OPRF.

Private E2EE
Sub-Linear Deep Learning

SLIDE Engine

Sub-LInear Deep learning Engine (SLIDE) uses Locally Sensitive Hashing (LSH) to achieve training and inference complexity that scales sub-linearly with network size.

L

50 Hash Tables

Multiple LSH tables for high recall candidate selection

K

6 Hash Bits

Per-table hash functions for bucket assignment

Adaptive Rebuild

Dynamic LSH index rebuilding with drift detection

Hash Family Support

SimHash Cosine Similarity
WtaHash Winner-Take-All
DwtaHash Dense WTA
MinHash Jaccard Similarity
SRP Sparse Random Projection
Performance

SLIDE achieves up to 3.5x speedup over TensorFlow on CPU for large output layers, with complexity O(LยทKยทC) instead of O(N) where C << N.

Private Inference

Secure two-party computation for confidential AI inference

Lower Security

~4.6 KB/token

Server learns nothing
Client embedding revealed
Plaintext forward pass
Fast execution
Best for
Low-latency applications with moderate privacy requirements

High Security

~2 MB/token, ~34K triples

Only norms revealed
Queries stay private
Gate values hidden
Slower execution
Best for
Maximum privacy for sensitive data and model protection

Cryptographic Primitives

F
Ferret OT
Correlated Oblivious Transfer
R
Ristretto255
OPRF Implementation
B
Beaver Triples
Multiplication Gates
Q
Fixed-Point
Q16.16 / Q32.32

LLM Inference

Production-ready LLaMA-compatible inference with optional sparsity

Architecture Support

GQA, RoPE, RMSNorm, SwiGLU - full LLaMA compatibility

Sparse Inference

50% head sparsity, 50% neuron sparsity with Deja Vu prediction

HuggingFace Compatible

Works with any LLaMA-architecture model in safetensors format

Inference Configuration

Parameter Default Description
temperature 0.7 Sampling temperature (0.0 = greedy)
top_k 40 Top-k filtering (0 = disabled)
top_p 0.9 Nucleus sampling (1.0 = disabled)
repetition_penalty 1.1 Penalize repeated tokens (1.0 = disabled)
max_new_tokens 512 Maximum tokens to generate
template auto Chat template (zephyr, chatml, llama2, llama3, mistral, raw)
Performance Verified

Benchmark Results

Criterion.rs performance analysis of klearu-core primitives

~68 ns
Sparse Dot Product
dim=1024, 10% density
~12.6 ยตs
SLIDE Forward Pass
128โ†’64โ†’10 network
~19.5 ยตs
Train Step
Forward + Backward
~128 ยตs
LSH Query Union
1000 neurons

Detailed Results

Benchmark Median Time Throughput Rating
Sparse dot product (dim=1024, 10% density) ~68 ns ~14.7M ops/sec Excellent
Sparse โ†’ Dense conversion (dim=1024) ~134 ns ~7.5M ops/sec Excellent
Dense โ†’ Sparse conversion (dim=1024) ~1.5 ยตs ~667K ops/sec Good
SimHash (dim=128, 4 tables) ~6.9 ยตs ~145K hashes/sec Good
SimHash (dim=1024, 4 tables) ~68 ยตs ~14.7K hashes/sec Expected
SLIDE forward pass (128โ†’64โ†’10) ~12.6 ยตs ~79K passes/sec Excellent
SLIDE train step (128โ†’64โ†’10) ~19.5 ยตs ~51K steps/sec Excellent
LSH query union (1000 neurons) ~128 ยตs ~7.8K queries/sec Acceptable

Star Result

~68 ns sparse dot product at 1024 dimensions with 10% density (~102 nonzero values). That's roughly 1.5 ns per multiply-add โ€” approaching peak SIMD throughput.

This is what enables sub-linear inference speed in SLIDE. The cache-aligned memory layouts in ContiguousWeightStore clearly pay off.

Optimization Target

LSH query at ~128 ยตs for 1000 neurons is the main bottleneck. At production scale with 100K+ neurons per layer, this dominates.

Use klearu-bolt autotuner to reduce K (hashes) and L (tables) to hit target recall while minimizing query cost.

Hash Function Scaling

SimHash dim=128 (4 tables) ~6.9 ยตs
SimHash dim=1024 (4 tables) ~68 ยตs

Scales linearly with dimension as expected (O(dim ร— K ร— L)). 8ร— dimension = ~10ร— time.

Recommendation: For high-dimensional dense vectors, prefer SRP (Sparse Random Projection) or DwtaHash for better throughput โ€” they skip zero entries.

Training Efficiency

~12.6 ยตs
Forward Pass

~79K passes/sec

~19.5 ยตs
Train Step

~51K steps/sec (+54% overhead)

Train step is only ~54% slower than forward pass โ€” the backward pass + Adam/SGD weight update adds very little overhead. Gradient computation stays within the sparse activated set.

Bottom Line

For a pure-CPU sparse deep learning engine, these numbers are solid and competitive. The core primitives (sparse dot product, forward pass) are well-optimized. The LSH query latency is the area to watch at scale โ€” use klearu-bolt autotuning to tune K/L and keep it in check. This is not trying to beat CUDA GEMM for dense workloads; it's targeting the sparse CPU inference niche, and within that niche it delivers.

Getting Started

Build and run Klearu in minutes

Installation

# Clone repository
git clone https://github.com/QuilibriumNetwork/klearu.git
# Build full workspace
cargo build --release
# Build with features
cargo build --release -p klearu --features full

Quick Start

# Download a model
huggingface-cli download HuggingFaceTB/SmolLM-135M-Instruct
# Run chat interface
cargo run --release --bin chat -- ./SmolLM-135M-Instruct
# Run diagnostics
cargo run --release --bin diagnose -- ./SmolLM-135M-Instruct
quilibrium_rankings.exe
Quilibrium
Holder Rankings
01
Black Hat
ELITE // ZERO-DAY ARCHITECT
3M+ ... wallets
02
White Hat
MASTER // ETHICAL EXPLOITER
1M-3M ... wallets
03
Grey Hat
VETERAN // SHADOW OPERATIVE
250K-1M ... wallets
04
Script Kiddie
OPERATOR // TOOL WIELDER
100K-250K ... wallets
05
Bug Hunter
SPECIALIST // VULN SEEKER
25K-100K ... wallets
06
Coder
DEVELOPER // BUILDER
5K-25K ... wallets
07
Learner
TRAINEE // STUDENT
1K-5K ... wallets
08
Noob
OBSERVER // GUEST
<1000 ... wallets