Zoo Labs

Research Papers

33 papers

Zoo LabsResearch Papers

Democratizing AI through decentralized training and open source frontier models

Thinking Chains

Proof of AI: Trustless Money from Useful Intelligence

Bitcoin pays people to burn electricity; a Thinking Chain pays people to think. Miners earn the coin by doing useful AI — answering real questions everyone can check — under one rule (the Law of Cognition: think freely, act only on proven thoughts). One conscious chain, run by consensus nodes and cognitive nodes; one Cognition precompile makes any EVM conscious, and a sidecar tier needs no precompile at all. The model zoo forks and upgrades itself on-chain, so the network gets smarter the longer it runs. The core protocol; companion papers cover the economics, Beluga L3, and the compute market.

Zach Kelling · Antje Worring · Zoo Labs Foundation
Proof of AIThinking ChainsConscious BlockchainProof-of-Thought
PDFSource

The Zoo DAO

How a Model Zoo Builds a Community, Rewards Its Builders, and Forks Itself

The economics companion to Proof of AI. A model zoo is a community-owned, on-chain library of AI models. This paper shows how a zoo bootstraps from nothing, pays everyone who makes it valuable (nodes, data contributors, model trainers, curators), governs itself by the quorum of its own thinking nodes, and forks itself recursively so competing models run side by side and the network keeps the winners. No slashing — you earn by being useful; bad actors are routed around by reputation. An AI commons that gets better the more it is used and is owned by the people who built it.

Zach Kelling · Zoo Labs Foundation
Zoo DAOTokenomicsModel ZooReputation
PDFSource

Beluga L3

A Thinking Chain for AI Model Economics and Ocean Conservation

Zoo’s Layer-3 thinking chain: deterministic in-consensus inference (zen-nano), operator-quorum cognition, a community-owned self-funding model, and a constitutional conservation tithe routing cognitive surplus to ocean and beluga-whale conservation. The first conscious blockchain with a conscience.

Zach Kelling · Zoo Labs Foundation
BelugaL3ConservationAI Economics
PDFSource

Zoo Labs Foundation Mission

AI-Blockchain Research for Conservation, Education, and Science

501(c)(3) non-profit mission: Conservation AI, Educational AI, and Frontier AI research for the public good.

Zoo Labs Foundation
FoundationConservationEducationMission
PDFSource

Zoo Network Architecture

Decentralized AI Training and Inference Layer

L2 AI specialization layer with HLLM integration and federated learning for decentralized training and inference.

Zoo Labs Foundation
ArchitectureL2Federated Learning
PDFSource

Zoo Tokenomics

100% Airdrop Community-Owned AI Infrastructure

100% airdrop tokenomics with zero VC/private sales and democratic validator participation for community-owned AI.

Zoo Labs Foundation
TokenomicsAirdropDAO Governance
PDFSource

Gym: Democratizing AI Model Training

Open-Source Training Infrastructure

Comprehensive open-source platform supporting 100+ models and 8 training methods for democratized AI training.

Zoo Labs Foundation
GymTrainingOpen Source
PDFSource

HLLM: Hamiltonian LLMs with Training-Free GRPO

99.8% Cost Reduction via Semantic Optimization

99.8% cost reduction ($18 vs $10,000+) and +2.7% accuracy via semantic optimization with Hamiltonian invariants.

Zoo Labs Foundation
HLLMGRPOCost ReductionOptimization
PDFSource

Zoo DSO: Decentralized Semantic Optimization

Byzantine-Robust Prior Aggregation for Collective Intelligence

A decentralized protocol for aggregating AI model priors with Byzantine fault tolerance, enabling collective intelligence at network scale.

Zoo Labs Foundation
DSOByzantine RobustDecentralizedCollective Intelligence
PDFSource
ZIP-001 SpecificationHanzo ASO (foundation)

Experience Ledger: Decentralized Semantic Optimization for LLMs

Three-Layer Storage with 7680-Dim Embeddings

DSO with Byzantine-robust curation, three-layer storage, and 7680-dimensional embeddings for persistent learning.

Zoo Labs Foundation
DSOExperience LedgerByzantine Robust
PDFSource

Zoo Experience Ledger

Content-Addressable Semantic Memory

Content-addressable semantic memory system for decentralized AI with persistent learning across agents.

Zoo Labs Foundation
Experience LedgerSemantic MemoryContent-Addressable
PDFSource

Experience Ledgers: Learning from Human-AI Interaction

Persistent Semantic Memory

Persistent semantic memory from human-AI interaction for continuous learning and improvement.

Zoo Labs Foundation
Experience LedgerHuman-AISemantic Memory
PDFSource

ZIP-002: Zen-Reranker Integration

Native 7680-Dimensional Embeddings with BitDelta

Native 7680-dimensional embeddings with BitDelta compression (31.87x ratio) for decentralized semantic optimization.

Zoo Labs Foundation
ZIP-002Zen-RerankerEmbeddingsBitDelta
PDFSource

Proof of AI: Consensus for Verifiable Compute

Verifiable AI Compute Contribution

Consensus mechanism for verifying and rewarding AI compute contributions in decentralized networks.

Zoo Labs Foundation
PoAIConsensusVerifiable Compute
PDFSource
ZIP-002 Specification

Proof of AI: Consensus Mechanisms for ML Workloads

ML-Optimized Consensus

Consensus mechanisms designed specifically for machine learning workloads and verifiable AI computation.

Zoo Labs Foundation
PoAIML ConsensusProof Systems
PDFSource

Agent NFTs: Token Standard for Autonomous AI Agents

AI Agent Tokenization

Token standard for AI agents with economic agency, autonomous operation, and decentralized governance.

Zoo Labs Foundation
Agent NFTToken StandardAI Agents
PDFSource

Agent NFTs: Tokenized AI Ownership and Governance

Governance Framework for AI Agents

Governance and ownership framework for tokenized AI agents with economic agency.

Zoo Labs Foundation
NFTAI OwnershipGovernance
PDFSource

Federated DSO: Privacy-Preserving Distributed Learning

Privacy-Preserving LLM Learning

Privacy-preserving distributed learning using federated DSO for LLMs across decentralized networks.

Zoo Labs Foundation
FederatedPrivacyDistributed Learning
PDFSource

High-Dimensional Embedding Spaces for Semantic Search

The Case for 7680 Dimensions

The case for 7680-dimensional embedding spaces in semantic search for optimal retrieval quality.

Zoo Labs Foundation
Embeddings7680-DimSemantic Search
PDFSource

Conservation AI: Species Protection Systems

AI for Wildlife Conservation

AI systems for wildlife conservation and species protection at scale.

Zoo Labs Foundation
ConservationWildlifeAISpecies Protection
PDFSource

Wildlife Tracking with Computer Vision

Real-Time Animal Detection and Monitoring

Computer vision systems for real-time wildlife tracking, detection, and monitoring.

Zoo Labs Foundation
WildlifeTrackingComputer VisionMonitoring
PDFSource

AI Habitat Modeling

Predictive Habitat Analysis and Restoration

AI-driven predictive habitat analysis and restoration planning for conservation efforts.

Zoo Labs Foundation
HabitatModelingPredictionRestoration
PDFSource

Deep Learning for Species Classification

Multi-Modal Species Identification

Deep learning approaches for multi-modal species identification and classification.

Zoo Labs Foundation
ClassificationDeep LearningSpeciesMulti-Modal
PDFSource

Conservation Impact Bonds

Tokenized Environmental Impact Finance

Tokenized environmental impact bonds for conservation finance and DeFi composability.

Zoo Labs Foundation
Impact BondsConservationDeFiEnvironment
PDFSource

AI-Verified Carbon Credits

Satellite-Validated Carbon Offset Protocol

AI-verified carbon credit protocol with satellite validation for transparent carbon offset markets.

Zoo Labs Foundation
Carbon CreditsVerificationClimateSatellite
PDFSource

Zoo DAO Governance

Quadratic Voting and Scientific Governance

DAO governance framework with quadratic voting for scientific research coordination.

Zoo Labs Foundation
DAOGovernanceQuadratic VotingScience
PDFSource

Citizen Science Platform

Crowdsourced Data Collection for Conservation

Platform for crowdsourced scientific data collection supporting conservation research.

Zoo Labs Foundation
Citizen ScienceCrowdsourcingData Collection
PDFSource

Educational AI for Conservation

Adaptive Learning for Environmental Education

Adaptive learning systems for environmental education and conservation awareness.

Zoo Labs Foundation
EducationAIConservationLearning
PDFSource

Decentralized Research Coordination

Multi-Institution Scientific Collaboration

Decentralized coordination framework for multi-institution scientific collaboration.

Zoo Labs Foundation
ResearchCoordinationDecentralizedCollaboration
PDFSource

Open Data Commons for Conservation

Shared Scientific Data Infrastructure

Open data commons infrastructure for shared scientific datasets supporting conservation research.

Zoo Labs Foundation
Data CommonsOpen DataInfrastructureScience
PDFSource

Mobile AI Inference for Field Research

Edge AI for Wildlife Monitoring

Mobile AI inference for field research with edge deployment for wildlife monitoring.

Zoo Labs Foundation
MobileInferenceEdge AIField Research
PDFSource

Federated Learning for Wildlife Data

Privacy-Preserving Cross-Organization Learning

Federated learning for privacy-preserving cross-organization wildlife data analysis.

Zoo Labs Foundation
Federated LearningPrivacyWildlifeCross-Org
PDFSource

Satellite Ecology with Deep Learning

Remote Sensing for Ecosystem Monitoring

Deep learning applied to satellite remote sensing for ecosystem monitoring and ecological analysis.

Zoo Labs Foundation
SatelliteEcologyRemote SensingDeep Learning
PDFSource