Framework-grade runtime for intelligence infrastructure

Build sovereign networks that run ownable intelligence.

Tetcore is a modular, deterministic execution framework that turns inference into protocol-native transactions: sharded model storage, relay-delivered prompts, receipt-based settlement, programmable revenue routing, and native contracts (TCL + TVM).

Native identity (Ed25519) No EVM • No Solidity Deterministic kernel Prunable storage ledgers

Why Tetcore

Traditional AI is centralized. Tetcore makes intelligence an infrastructure primitive: composable, auditable, economically routable, and deployable as sovereign networks.

Ownable shards & models

Asset-grade intelligence

Register models by commitment. Split weights into shards with explicit ownership, policy hooks, and revenue routing.

Private by design

No prompt on-chain

Prompts stay local. Only commitments and receipts reach consensus. Relay transport is encrypted end-to-end.

Deterministic settlement

Receipts, not trust

Inference is off-chain; settlement is on-chain. Metered receipts and policy enforcement keep economics honest.

Architecture

A framework-style core that composes modules into a single deterministic state machine.

Protocol Kernel Δ(S,B)

Deterministic state transitions

Canonical transaction validation, module dispatch, receipt verification, and atomic revenue routing.


State Accounts • Models • Policies • Nodes • Receipts • Vaults • Contracts • Governance

Multi-ledger design

Prunable storage ledgers

Core commitments are permanent. Weight shards live on storage ledgers with TTL + challenge windows + availability checkpoints.


Goal high throughput without chain bloat.

Nodes selectable modes

Client • Validator • Inference • Relay

Run only what you need. Validators secure consensus. Operators serve inference. Relays transport prompts privately.

Inference as transactions

Prompt → Receipt → Settlement

SubmitPrompt locks escrow. Operators stream output. SubmitReceipt triggers deterministic distribution by policy.

Governance built-in

Upgrades with timelocks

Parameter updates, capability governance, and safe protocol upgrades via typed proposals and deterministic execution.

TCL + TVM

A native smart contract language and deterministic VM built for infrastructure-grade safety and capability-based authority.

TCL strong typing

Capability-first contracts

Contracts declare the exact capabilities they need. No ambient authority. No floating point. Deterministic ABI.


Example pricing policies, revenue routing, vault logic, governance extensions.

TVM gas-metered

Deterministic execution

Register-based VM, bounded memory, explicit gas accounting, storage commits only on success.


Rule same input + same state ⇒ same output on every node.

Economics

Fees are workload-sensitive and routed on-chain. Users can stake into models for revenue participation without securing consensus.

Pricing owner / market / hybrid

Policy-driven fees

Owner-set formulas, market clearing, or bounded hybrid. Fees scale with prompt/output tokens and complexity.

Routing basis points

Automatic distribution

Split fees between inference operators, model owners, shard providers, validators, and vault participants.

Vaults model-level staking

Stake to participate

Deposit network tokens into a model vault (if enabled by owner) to receive proportional revenue share.

Bootstrap a Network

A “framework-grade” developer workflow: compose modules, generate genesis, run nodes, iterate.

Quickstart local devnet
# 1) Install toolchain cargo build # 2) Generate a blueprint + genesis tetcore-sdk blueprint init --name devnet tetcore-sdk genesis build --blueprint devnet.json # 3) Run node modes tetcore-node --mode validator --genesis genesis.json tetcore-node --mode relay tetcore-node --mode inference --model default-mini # 4) Submit prompt (local agent) tetcore-client prompt --model default-mini --text "hello" --fee-limit 0.02
Open Docs