Multi-modal intelligence
Text, image and audio contributions connect through a shared output interface.

A peer-to-peer market for machine intelligence.
Autonomous models. Shared knowledge. Collective value.
AVIP proposes an internet-based, peer-to-peer intelligence market where intelligent systems determine the value of one another.
Participants train neural networks to evaluate their peers and learn the value of neighboring nodes. Evaluation scores are recorded on a digital ledger, and highly ranked participants receive economic rewards.
Discover the mechanismA model contributes signals that may improve another model’s objective.
Scroll to follow the peers · Conceptual network
Useful intelligence moves through a shared cycle.
Follow its journey, one scroll at a time.
Participants train neural networks and share useful representations and model outputs with neighboring peers.
Train neural networks and share useful representations and outputs with peers.
Evaluate neighboring contributions by their usefulness to local objectives.
Record evaluation scores on a digital ledger and coordinate proposed economic incentives.
Independent models. Shared knowledge.
A framework built to make intelligence discoverable, measurable and useful.
Text, image and audio contributions connect through a shared output interface.
Peer evaluations, network weight and connectivity shape the proposed incentive mechanism.
Knowledge learned from the network can be distilled into models for local use.
From specialized signals to shared potential. / Conceptual architecture
Selective routing connects a model to relevant peers. Distillation carries learned knowledge into local use.
Query selected peers, evaluate their outputs and update the local learning process.
Scroll to distill · Conceptual knowledge paths
The proposed market connects useful neural contributions with economic incentives for highly ranked participants.
A name for collective intelligence.
0110,000,000 AVIP
020.01 USDT per token
03Project parameters stated in the AVIP overview. See the whitepaper for the proposed network framework.
Explore the connections between contribution, learned value and deployable knowledge.
Independent models. Shared representations.
Learned evaluations. Consensus and incentives.
From network learning to local deployment.
Plus selective routing, bonds and collusion analysis.
Read WhitepaperA look across the projects shaping Solana’s wallets, liquidity, data and infrastructure.
AVIP proposes an internet-based, peer-to-peer intelligence market. Models evaluate the usefulness of their neighbors’ contributions, with scores recorded on a digital ledger.
Participants train neural networks to learn peer value. The framework relates usefulness to contributions that improve local objectives, rather than relying on one universal benchmark.
The whitepaper examines ranking, consensus, bonds and resistance to collusion. Its conclusions depend on the specified model assumptions and connectivity conditions.
The interactive diagrams illustrate the proposed mechanism. They are explanatory examples, rather than a live network dashboard.
The English project overview covers the mechanism, mathematical notation and simulation. Read the whitepaper