DevNet6 is launching with Cortensor’s first full-stack deployment, marking the shift from modular experimentation to cohesive protocol validation. It brings together adaptive miner coordination, enforced payments, dual testnets, and the infrastructure to scale into a permissionless execution layer.
Status: Launching with DevNet6 Purpose: Dynamically classify miner nodes based on:
Historical performance
Token throughput
Inference quality (ack → precommit → commit)
Enables:
Tiered prompt complexity and task allocation by node capability
Hardware-aware job routing (from Raspberry Pi → H100)
Support for future multi-model deployment
Significance: Moves Cortensor beyond “one-model-fits-all” toward adaptive, scalable AI coordination across a heterogeneous, decentralized network.
Core priorities for DevNet6:
Validate prompt compatibility across Cognitive Levels
Refine task scoring logic (ack → precommit → commit weighting)
Enforce SessionPayment credit and payout mechanics
Iterate miner/oracle sync to keep CognitiveLevel accurate
Improve UI/dashboard for session funding and payment visibility
Begin validator set planning for oracle rotation
Explore practical privacy integrations:
Session-based authentication
Diffie-Hellman key exchange
Optional Trusted Execution Environment (TEE) support
Stable environment for app developers
Used for validating session workflows
Primary testnet for public onboarding and dApp integration
A dedicated rollup for modeling Cortensor’s gas/token economy using a staged rollout:
Phase 1: ETH-as-Gas (DevNet6 & Testnet Phase)
Default gas token during testing
Enables rapid iteration with familiar infra
No need for COR token faucet mechanics
Phase 2: COR-as-Gas (Post-Mainnet Upgrade)
Enables COR to become true fuel for computation
Creates circular economic flow between users, miners, and governance
Provides deflationary pressure via gas usage
Activated via rollup config without contract redeploys
DevNet6 tests infrastructure. Testnet validates coordination. Mainnet unlocks economic flywheel.
Hardware-aware classification enables multi-model routing starting in DevNet6.
Initial Model Targets:
DeepSeek
Qwen
Other open-weight LLMs under review
Next Steps:
Prompt tuning and miner benchmarking
Integration at task level
Gradual scaling across network
DevNet6 begins a continuous cycle of internal and external hackathons to mature the ecosystem:
Target early tooling, dashboards, bots, SDK flows
Drive improvements in observability, job tracing, and session quality
Will expand as E2E stack stabilizes
Focus on agent apps, frontends, wallets, and composable use cases
Rewards and grants tied to long-term utility
Community Tools & Collaboration:
📖 References:
Final internal environment
Dataset expansion, coordination under pressure, and validator refinement
Used to evaluate live network behavior and validator performance
Will guide validator set selection and final mainnet adjustments
Will follow successful public Testnet
COR-as-Gas and full token utility enabled
Validator set, reward logic, and governance modules activated
Last updated
