QuantumAI BlockchainPost-quantum Layer 1 · EVM chain ID 3303

The trust layer for
artificial intelligence.

QuantumAI Blockchain is a live post-quantum network where an AI's training, evaluation, attestation and rewards are consensus, not the private decision of one company. Nobody owns the model. The network does.

Trains

Any node can contribute compute; verified epochs finalise on-chain.

Verifies

Model identity and reasoning are attested by validator quorum.

Serves

The attested model answers at aethersmind.com, checkable against the chain.

Aether model

Syncing

Knowledge vectors

Syncing

QBC block

Syncing

Network nodes

Syncing

Φ (experimental)

Syncing

Supply

Syncing

Aether helps people think, build and understand.

Use Aether for research, explanation, writing, coding, structured analysis and difficult technical work.

Research and explanation

Synthesise sources and explain hard ideas clearly.

Writing and structured thinking

Draft, outline and reason through structured problems.

Technical analysis

Analyse systems, trade-offs and failure modes.

Coding and system design

Design, review and debug software and architecture.

Powerful AI should not depend on one private decision-maker.

QBC turns model evolution into a verifiable distributed-systems problem: it coordinates identity, model-state commitments, validator attestation, smart-contract execution and rewards.

Conventional system

  • ·Company owns model
  • ·Company chooses training
  • ·Company evaluates privately
  • ·Company selects checkpoint
  • ·Users receive the result

Aether and QBC

  • +Workers propose improvements
  • +Validators reproduce measurements
  • +Quorum attests the checkpoint
  • +Canonical model state advances
  • +The network records and rewards contribution

Train. Evaluate. Attest. Finalise. Reward.

A candidate update does not become canonical because one operator reports a good result. It must pass the protocol's measurement and attestation process.

01Live

Train

A signed worker creates a candidate update from the current canonical model state.

02Live

Evaluate

Held-out evaluation tests the candidate on material excluded from training.

03Private network

Attest

Validators independently reproduce the agreed evaluation.

04Live

Finalise

The canonical model root advances only after the protocol's quorum and state-transition requirements are satisfied.

05Live

Reward

Accepted contributions produce protocol settlement and auditable reward records.

Measured, not implied.

Inspect the model identity, benchmark settings, sample scope, baselines, result artifacts and complete model lineage.

aether-v7.1-unified

Frozen Qwen2.5-7B-Instruct base · 10-expert Sephirot adapter (rank 16, top-2)

Live
MMLU (adapter)
71.17%
MMLU baseline
71.28%
GSM8K (adapter)
77.8%
GSM8K baseline
67.8%+10.0 pp
MMLU samples
14,042
GSM8K samples
625
Harness
aether-v7-eval
Base parameters
7B

Watch the network train Aether.

Follow active rounds, signed worker submissions, held-out evaluation, validator attestations, checkpoint finalization and reward settlement.

Network state

  • Consensus and attestationLive
  • Single-worker training loopLive
  • Multi-node training designTested
  • Public cross-worker aggregationActivation pending
  • Signed gradient protectionLive

The evidence is public.

Explorer

The QBC Explorer is the transaction-level source of truth: blocks, transactions, contracts, addresses, validator actions, model attestations and training rewards.

Open Explorer →

Dashboard

The Dashboard is a system-wide operating view of QBC nodes, validators, chain health, Aether services, Aether-Net and QVM activity.

Open Dashboard →

Deploy smart contracts on QBC.

Build Solidity applications on QBC's EVM-compatible execution layer and connect them to the network's native infrastructure.

Chain ID 3303Block time ~3.3sEVM-compatible tooling

Help operate the network.

Run validation infrastructure, contribute training work or prepare to serve distributed inference.

Validator

Live

Reproduce checkpoint evaluations, participate in network consensus and help determine canonical protocol state.

Run a validator →

Training operator

Private network

Produce signed candidate updates and participate in Aether's model-improvement process.

Train Aether →

Inference operator

Activation pending

Future distributed inference providers will serve complete Aether replicas and compete on availability, latency and throughput.

Join inference updates →

One model. Inspectable lineage. Network coordination.

Aether inference runs in the Aether runtime. QBC records and enforces model identity, checkpoint state, attestations, policy events and settlement.

Frozen base + adapters

A frozen Qwen2.5-7B base with Sephirot routing adapters, with capability preserved and evolution isolated to the adapter.

Knowledge Fabric

Provenance-linked retrieval vectors across ten cognitive domains, with reranked search. Long-term scale target: one trillion provenance-linked vectors.

Checkpoint identity

A model-card identity and checkpoint root recorded on QBC, with attention-derived Φ reported as an experimental signal.

Inspect the evidence.

Read the papers, examine the model card, review the evaluation harness and reproduce the published results.