Cortensor Announcements

🧭 Mainnet Full Alpha Is Kicking Off

Mainnet Full Alpha marks the next stage of the Cortensor rollout.

This is not a one-time launch event. Alpha is the period where Mainnet1, the product layer, and developer-facing tools start being exercised together under more realistic usage before we expand further toward Beta.

🔹 What Alpha is for
- controlled access rather than broad public availability
- continued validation of Mainnet1, routers, sessions, dedicated nodes, dashboard, indexer, and recovery paths
- real product traffic through Portal, Corgent, Bardiel, and PyClaw-related workflows
- identifying operational, compatibility, onboarding, model, and capacity gaps through actual use

🔹 Product focus
- Portal will continue improving as the simpler hosted API entry point into Cortensor
- Corgent / Bardiel will keep iterating on delegation, validation, fact-checking, and the early v4 direction
- PyClaw will continue maturing across research, coding, memory, tools, scheduling, delegation, recovery, longer-running projects, packaging, and the early Desktop App
- Mainnet Full capacity will gradually expand as we learn where real demand appears

🔹 Developer exposure
- Alpha is also preparation for broader builder activity
- We expect to prepare more examples, documentation, onboarding, model availability, and product capacity for hackathons and builder programs
- Portal and PyClaw should provide two accessible entry points, while Corgent / Bardiel can give agents additional delegation and validation services to build around

🔹 What happens between Alpha and Beta
- bring in controlled real usage
- expose gaps through actual workloads
- fix bugs and compatibility issues
- improve onboarding and developer experience
- mature PyClaw and the product surfaces
- iterate v4 delegation / validation
- expand node, model, router, and gateway capacity
- run early hackathons / builder programs
- use that feedback to prepare for broader Beta access

🔹 What Beta should represent
Beta should mean more than simply “more users.”

It should mean a stronger network and product baseline that has already gone through real developer usage, more diverse workloads, broader model testing, and another round of operational hardening.

Alpha → real usage → iteration → developer exposure → more workloads → stronger products → Beta.

Mainnet Full Dashboard:
https://dashboard.cortensor.network

Mainnet Full Explorer:
https://explorer.cortensor.network

https://x.com/cortensor/status/2107074852934852657

🧪 Testnet Phase #4 – Wrap-Up Recap

Testnet Phase #4 is wrapping up over the next 48 hours.

Phase #4 focused on tightening the full Cortensor stack under more realistic conditions - network execution, Portal, router pools, observability, L3 recovery, and Mainnet Lite readiness.

Core Infra:
🔹 Mainnet Lite baseline validation
- Mainnet Lite moved from rough setup into real E2E baseline checks under actual mainnet conditions
- Dedicated-node user tasks, ephemeral-node network tasks, and ephemeral-node user-task → node-pool flows all worked end to end

🔹 Mainnet Lite infra readiness
- Contract/module deployment, RPC/server setup, oracle/indexer/dashboard setup, explorer checks, and chain-side validation reached rough baseline form
- Heartbeat/ping timing was adjusted around real Arbitrum mainnet gas-cost assumptions

🔹 RPC / environment separation
- Internal RPC infrastructure was refreshed and migrated toward newer internal/community-backed paths
- Testnet and mainnet RPC separation started on newly provisioned servers as prep for Mainnet Lite and later Mainnet Full

🔹 Testnet1a / L3 recovery
- Testnet1a went through a real sequencer / local Nitro DB recovery exercise using L2 batch data plus DAS/DA-backed data
- The recovery helped validate L3 rebuild flow, edge cases, and future sequencer backup/redundancy planning

🔹 Cortensor L3 progress
- Cortensor L3 closed Phase #4 in a much stronger operational position
- The network has processed over 8M blocks and over 16.5M transactions

🔹 Mainnet Full early tryouts
- Very early Mainnet Full infra work started across rough server setup, bridge flow, contract deployment, explorer checks, and connectivity validation
- This remains early, but helped prepare the path beyond Mainnet Lite

Products / Stack:
🔹 Portal V1 – hosted API layer
- Portal stress tests expanded into larger prompts, heavier token-size requests, longer-running hosted API calls, and multi-model testing across oss-20b, gemma4, qwen 3.5 9b, and oss-120b
- API Gateway → router pool → session routing was validated in rough form, including dynamic model reconfiguration, capacity-aware routing, and multi-instance gateway deployment

🔹 Portal V1 – metering + observability
- OpenAI-style and Anthropic-style REST compatibility advanced, with SSE / streaming MVP support added at the gateway level
- Usage tracking moved toward token-based metering, while public stats and admin/ops views expanded across usage, latency, route health, gateway health, tokens, heatmaps, and model/session triage

🔹 Portal auth + control-plane direction
- Portal auth matured with prod OAuth setup for Google, X, and Discord, while wallet login was de-emphasized because of spam/abuse concerns
- Backend planning advanced toward a future API/control-plane path beyond the current Supabase-style pattern

🔹 Docs V2
- Docs V2 progressed in rough form with cleaner structure, updated diagrams, and clearer operator/developer paths
- Current direction is to preview the new docs website during Q3 between Mainnet Lite and Mainnet Full

🔹 PyClaw
- PyClaw early WIP direction continued with a public early snapshot/repo push
- Framing became clearer: local-first agent/app layer now, with future connection into remote delegation and agent-cloud-like flows

🔹 Stack messaging
- Public stack messaging expanded across Network & Infra, Dashboard, Portal, Corgent, @BardielTech, and PyClaw
- Mainnet Lite vs Mainnet Full was clarified further: Lite first as the controlled L2 rollout path, Full later as the fuller Arbitrum Orbit L3 path

Phase #4 tightened the path from network execution to hosted access, routing, metering, observability, L3 recovery, and Mainnet Lite readiness.

After Phase #4 wraps, the next focus is detailed planning and execution toward Mainnet Lite.

https://x.com/cortensor/status/2073337304278843903

🧪 Testnet Phase #4 Starts Now

Testnet Phase #4 is now live.

Phase #4 is less about adding isolated features and more about tightening the full pipeline under realistic conditions — data → quality → routing, reliability under load, and Mainnet Lite readiness.

🔹 E2E scope lock (Phase #3 → Phase #4)
Finalize the end-to-end test plan + order + pass/fail metrics (throughput/latency, recovery, drift, data integrity/offchain correctness)

🔹 Deeper stress testing (Data → Quality Path)
Longer, higher-volume runs that hit real failure modes (miner churn, partial outages, RPC degradation, queue spikes)

🔹 Privacy + offchain storage stress tests (optional / heavier)
Heavier testing where enabled: encryption paths, offchain correctness, and edge cases (timeouts, stale keys, partial writes)

🔹 SLA / quality signals – deeper validation (#3 signals)
Stress SLA signals under churn/load (drift detection, reliability weighting vs uptime, disagreement/arbitration behavior)

🔹 Portal V1 (design → early implementation) + product boundary
Portal becomes a real product layer: client → Portal backend → router pool → sessions
Scope: API keys, usage visibility, managed router pools, auth/rate limits, billing hooks + metering

🔹 Portal focus (gateway → router pool path)
Key issuance/consumption path is proven; Phase #4 drills into the hosted execution shape from Portal Gateway into managed router pools

🔹 PyClaw (early dev release track)
Move from design into early dev release: landing page + docs + initial runtime scaffolding aligned to router/session primitives

🔹 Mainnet Lite infra planning (execution readiness)
New RPC/server path is ready (L2 testnet/mainnet + L3 testnet/pre-mainnet)
Phase #4 begins bringing up Mainnet Lite pieces on top (dashboard, indexer, oracle-related path)

🔹 Core Docs
Testnet Phase #4: https://docs.cortensor.network/community-and-ecosystem/community-testing/testnet-phase-4

https://x.com/cortensor/status/2055830240958533777

🧪 Testnet Phase #3 – Wrap-Up Recap

Testnet Phase #3 has been running for a few weeks, and we’re now wrapping up the current Phase #3 stretch. We moved from setup → real testing loops → product/data-layer alignment, with v3 agent surfaces + MVP data management + inference-quality signals exercised together.

Core Infra:
🔹 v3 /delegate + /validate (testing + iteration)
- Rollout + session prep complete; now iterating on real delegation/validation behavior
- Focus: routing consistency, consensus handling, and remaining logic gaps under load

🔹 MVP Data Management (Privacy + Offchain)
- Privacy Feature 1.0 + Offchain Storage v3 working end-to-end in MVP form
- Includes encryption (session/task), dedicated/ephemeral flows, deferred write, router→miner→dashboard integration

🔹 Dashboard / UI / UX improvements
- Quality stats + privacy/offchain flows now surfaced more clearly in the dashboard
- Added rank table + Node Perf quality columns + clearer completed/processing/stale task views

🔹 Inference Quality (SLA #3 live path)
- SLA #3 (user-task quality stats) added as a third selection filter on top of node + network-task signals
- Deployed on testnet0 + testnet1a and now being observed in live selection behavior

🔹 Inference Quality (oracle + data layer)
- MVP Quality Oracle + data model working end-to-end
- Task probes → stored stats → ranking views + selection/reward-oriented surfaces

🔹 Privacy Feature 1.0 (testing stage)
- Private inference live across router surfaces, tested across dedicated + ephemeral paths
- Session-level + task-level encryption flows have gone through matrix-style testing

🔹 Offchain Storage v3 (usable MVP path)
- Dedicated + ephemeral offchain flows working in MVP form
- Router-assisted deferred-write path for ephemeral nodes is live (remaining gaps tracked for iteration)

Products:
🔹 Bardiel dashboard – v3 adaptation
- Alignment work underway for v3 /delegate + /validate flows
- Heavier datasets + API examples added; iterating on task cards, results, consensus views, longer-input UX

🔹 New models for PyClaw iteration
- Added models 73–77 (Gemma 4 + Qwen variants) for PyClaw experiments
- Focus: tool-calling gaps and agent-style behavior on newer OSS models

🔹 PyClaw (early iteration + alignment)
- Early experiments started on newer models with dedicated sessions
- Focus: runtime/agent gaps + prep for deeper experimentation next phase

🔹 Portal V1 – design scoping started
- Scoping Portal V1 as a hosted inference layer on top of router infrastructure
- Focus: API key access, usage visibility, and managed router pools for developers

🔹 Router Node – multi-session + load balancing work
- Started design/code work for multi-session routing + load balancing on router nodes
- Supports Portal + PyClaw use cases with better traffic distribution and reliability

Phase #3 connected the pieces: agent surfaces (v3) + data management (privacy/offchain) + quality signals (SLA #3). Next is tightening reliability, UX, and mainnet-readiness based on what Phase #3 surfaced.

https://x.com/cortensor/status/2050822142485569685

🧪 Testnet Phase #3 – Progress Recap So Far (1–2 Weeks Before Wrap-Up)

Testnet Phase #3 has been running for a few weeks now.

So far, Phase #3 has moved from initial setup → real testing loops → early product/data-layer alignment. We now have v3 agent surfaces, MVP data management, and inference-quality signals all actively exercised together.

🔹 v3 /delegate + /validate (prep → testing phase)
- Initial rollout + session prep is done.
- We’re now in deeper testing + iteration on real delegation/validation behavior, routing consistency, consensus handling, and remaining logic gaps.

🔹 MVP Data Management (now fully in place)
Privacy Feature 1.0 + Offchain Storage v3 are working end-to-end in MVP form: session/task encryption, dedicated/ephemeral flows, deferred write, and router → miner → dashboard integration.

🔹 Dashboard / UI / UX (closing the gap)
Quality stats, privacy flows, offchain behavior, and task-status improvements are now surfaced in the dashboards - including the quality stats rank table, Node Perf quality columns, and clearer completed / processing / stale task views.

🔹 Inference Quality (SLA #3 live path)
- NodePool + NodePoolUtils now include SLA #3 (user-task quality stats) on top of node-level + network-task signals.
- The new quality path is deployed on testnet0 + testnet1a and is being observed in live selection behavior.

🔹 Inference Quality (oracle + data layer)
- Early E2E Quality Oracle + data model is working in MVP form.
- Real task probes now flow into stored quality stats, ranking views, and selection/reward-oriented surfaces.

🔹 Privacy Feature 1.0 (ready → testing stage)
- Private inference is now live across the router surfaces and tested across dedicated + ephemeral paths.
- Session-level and task-level encryption flows have both gone through matrix-style testing.

🔹 Offchain Storage v3 (usable MVP path)
Dedicated-node and ephemeral-node offchain storage flows are both working in MVP form, including the router-assisted deferred-write path for ephemeral nodes.

🔹 Bardiel dashboard – v3 adaptation started
- Dashboard iteration is now actively underway to align Bardiel with v3 /delegate + /validate flows.
- We’ve generated larger and heavier test datasets, added API examples, and started refining task cards, result rendering, consensus views, and longer-input UX.

🔹 New models for PyClaw iteration
Added newer models (Gemma 4 + Qwen variants, model IDs 73–77), built images, updated runtime/dashboard support, and enabled selected Gemma 4 models on ephemeral-node runtime for near-term experiments.

🔹 PyClaw (early iteration + alignment)
- Initial iteration is starting around newer Gemma 4 models on dedicated sessions.
- Focus is still on identifying agent/tool-calling gaps on newer open models and preparing for further experimentation.

Phase #3 so far has successfully connected the pieces: agent surfaces (v3) + data management (privacy/offchain) + quality signals (SLA #3).

The remaining stretch is less about adding missing foundations and more about testing, refinement, and tightening how these pieces behave together under real conditions.

https://x.com/cortensor/status/2048326864494829666

🧪 Testnet Phase #3 Starts Now

Testnet Phase #3 is now live.

Phase #3 pushes Cortensor from “trust-layer hardening” into economic + agentic hardening - payments/staking become production-shaped, rewards become more repeatable, and router surfaces move closer to agent primitives.

🔹 Privacy feature v0.5
Router key distribution (ACL/signature checks) + dashboard privacy module + dedicated sessions marked private/public

🔹 Payments (SessionPayment “production shape”)
Tune unit costs/rebates/surcharges + expand pricing metadata + improve cost attribution + failure handling

🔹 Dynamic pricing activation
SLA / depth / redundancy-aware fees via SessionPaymentTable + consistency + manipulation-resistance checks

🔹 Staking-to-use
Quotas → auto-deduction + usage buckets + cadence + anti-abuse guardrails (dashboard-visible)

🔹 Rewards (semi-automation)
Push reward outputs (CSV) further end-to-end to remove more manual steps (human still final send/batching) + keep tuning level params (capacity/breadth)

🔹 Router surfaces (agent-ready)
Iterate /delegate + /validate v3/v4 with policy hints, tool traces, retries/fallbacks + explicit redundancy/consensus attributes + expand factcheck/new endpoints

🔹 Bardiel dashboard + regular test jobs
Keep views stable across schema changes + scheduled smoke/regression jobs so examples stay fresh

🔹 Pre-mainnet experiments
Mainnet-identical dry runs: ops/reliability, cost mechanics, payment/staking/reward flows, agent-facing behavior

🔹 Self-managed L3 as primary track (Testnet-1a)
Ops hardening + runbooks (backups, snapshots, disk estimates) + validate procedures under real incidents

🔹 ERC8004 agent workflows
Corgent + Bardiel iteration via router MCP/prototype stack (real agent-facing trust rails)

🔹 Gas profiling iteration
Measure deltas vs Phase #2 baseline + keep trimming writes/log spam for mainnet viability

🔹 PyClaw (router-aligned local agent runtime)
Phase #3 = design + alignment for Phase #4 implementation/dogfooding

🔹 Core Docs
Testnet Phase #3: https://docs.cortensor.network/community-and-ecosystem/community-testing/testnet-phase-3

PyClaw – local-first agent runtime design: https://docs.cortensor.network/community-and-ecosystem/products-and-agents/pyclaw

/delegate & /validate – next router-node agentic surface
- v3: https://docs.cortensor.network/community-and-ecosystem/products-and-agents/corgent/delegate-and-validate-v3-spec-explicit-redundancy-and-consensus
- v4: https://docs.cortensor.network/community-and-ecosystem/products-and-agents/corgent/delegate-and-validate-v4-spec-programmable-consensus-and-receipts

Private / Encrypted Inference v0.5: https://docs.cortensor.network/technical-architecture/data-management/private-encrypted-inference/v0.5-contract-backed-acl-for-dedicated-private-sessions-session-based-only

https://x.com/cortensor/status/2033399448152490114

🧪 Testnet Phase #2 — Wrap-Up Recap (Rewards Complete)

Testnet Phase #2 is now officially wrapped - and reward/prize distribution is complete.

Phase #2 delivered what we wanted: a stable baseline, meaningful self-managed L3 progress, and a stronger trust + routing foundation to support the more agentic direction of Phase #3.

✅ What We Completed in Phase #2

🔹 Stable baseline achieved
- Network/background tasks held steady enough that Phase #2 stats became real signal (usable for routing, not noise)

🔹 Self-managed COR L3 is live as a real track (Testnet-1a / Arbitrum Orbit)
- Completed apple-to-apple L3 comparison and stood up Testnet-1a as the self-managed COR L3 environment
- Shifted migration + load-testing weight toward self-managed L3 so results are actionable

🔹 Routing filters upgraded using real metrics
- Moved toward “earned routing priority” using reliability/behavior/consistency signals
- Added QuantitativeStats + QualitativeStats as secondary filters to improve routing quality beyond simple uptime

🔹 Validator engine experiments rolling out (#Ollama + embeddings)
- Started deploying Ollama judge models + multiple embedding variants
- Reduced environment-specific validator issues while improving scoring consistency + throughput across mixed CPU/GPU setups

🔹 Agent-facing router surfaces expanded
- Live + exercised: /delegate v1/v2, /validate v1/v2, and /factcheck
- Bardiel dashboard now has mock/example datasets for agent-side iteration (more real data + refinement coming)

🔹 ERC-8004 mainnet agents registered (Corgent + Bardiel)
• Corgent + Bardiel are now registered as ERC-8004 mainnet agents backed by the Router Node MCP/prototype stack

🔹 Self-managed L3 ops playbooks completed
- Produced ops playbooks for self-managed L3: backups, snapshots, recovery flow, disk space estimates

🔹 Pre-mainnet environments progressed
- Pre-mainnet setup/env work is in place - deeper mainnet-identical experiments land in Phase #3

📌 Where Phase #3 Goes From Here

Phase #3 is more agentic + router-node iteration: we’ll keep tightening /delegate + /validate surfaces (and likely add more agent-focused endpoints), and we’ll run pre-mainnet experiments on the self-managed L3 setup designed to be identical to mainnet.

Thanks to all node operators and testers who made Phase #2 productive end-to-end.

https://x.com/cortensor/status/2028040657877815459

🧪 Testnet Phase #2 - Recap

Phase #2 delivered a stable baseline, real traction on the self-managed L3 track, and measurable progress on the trust layer (PoI/PoUW) + observability - plus the groundwork for more agent-ready router surfaces going into Phase #3.

✅ What We Completed in Phase #2

🔹 Network stability held as a usable baseline
- Background/network tasks stayed stable enough that Phase #2 stats became real signal (good enough to drive routing decisions)

🔹 Self-managed COR L3 is now live as a real track (Testnet-1a / @Arbitrum Orbit)
- Completed the apple-to-apple L3 comparison and stood up Testnet-1a as the self-managed $COR L3 environment
- Migration + load-testing momentum shifted toward self-managed L3 so results aren’t theoretical

🔹 Routing filters upgraded
- Added QuantitativeStats + QualitativeStats as secondary filters to improve reliability/behavior/consistency weighting (not just uptime)

🔹 Validator engine experiments shipped and rolling out (#Ollama + embeddings)
- Started rolling out Ollama judge models + multiple embedding variants
- Reduced environment-specific issues while improving scoring consistency + throughput across mixed CPU/GPU setups

🔹 Router Node surfaces expanded for agent workflows
- Live + exercised: /delegate v1/v2, /validate v1/v2, and /factcheck
- @BardielTech dashboard now includes mock/example datasets to validate delegation/validation UX end-to-end (more iteration next phases)

🔹 #ERC8004 mainnet presence established (Corgent + Bardiel registered)
- Corgent + Bardiel are now registered as #ERC8004 mainnet agents backed by the Router Node MCP/prototype stack

🔹 L3 ops hardening + playbooks completed
- Built an ops playbook for self-managed L3 including backups, snapshots, and disk space estimates (plus recovery-oriented runbooks)

🔹 Pre-mainnet environments progressed
- Pre-mainnet setups/environments are now in place (deeper dry-run experiments next)

📌 What Phase #3 Will Focus On

🔹 More agent-ready Router Node iteration
Phase #3 shifts more toward agentic readiness - tightening and expanding the router surface so agents have better primitives to build on (more iteration on /delegate, /validate, factcheck, and likely additional endpoints needed for agent workflows).

🔹 Commit to the self-managed L3 path
We’re now aligning around the self-managed $COR #L3 @Arbitrum Orbit track as the forward path. Phase #3 will treat this as the primary environment for deeper testing.

🔹 Pre-mainnet experiments (mainnet-identical setup)
During Phase #3, we’ll run pre-mainnet experiments designed to be identical to mainnet configuration, so we can validate costs, ops, reliability, and agent-facing flows under realistic conditions before going live.

https://x.com/cortensor/status/2026924572700774613

🧪 Testnet Phase #2 Starts Now

Phase #2 is live - this round is about validation refinement + infrastructure hardening, so Cortensor’s trust layer becomes more deterministic, more redundant, and cheaper to run.

All environments are now active and wired for Phase #2 testing:
- Testnet-0 — #Arbitrum #Sepolia (L2)
- Testnet-1 — $COR #L3 via RaaS
- Testnet-1a — Self-managed $COR #L3 (comparison target)
- Testnet-1b — Self-managed $COR #L3 (internal dev environment)

🔹 Focus – Validation Refinement (PoI + PoUW)
- Tighten Validator v2/v3 scoring so outputs are more deterministic, less “flaky,” and better rubric-aligned.
- Expand redundancy + aggregation testing so trust stays stable under disagreement and partial failures.

🔹 Validator Engine Experiments – #Ollama + Embeddings
- Trial Ollama judge models and multiple embedding variants so validators run reliably across mixed CPU/GPU hardware.
- Reduce validator environment-specific issues while improving scoring consistency and throughput.

🔹 /delegate & /validate Surface Hardening
- Exercise agent-facing flows: policy hints, multi-run delegation, retries/fallbacks, and validation-heavy workloads.
- Confirm response structure stability for downstream apps/agents.

🔹 Infra + Service Reliability
Verify router APIs, indexers, and dashboard feeds stay fresh and consistent under sustained load and recovery scenarios.
Add and test RPC redundancy, failover, and validator backups so the network survives infra hiccups gracefully.

🔹 Gas Optimization & Runtime Tuning
- Profile cortensord + L2 execution paths.
- Optimize SessionQueueValidation, SessionPayment, and validator storage to trim gas + runtime overhead.

🔹 Dashboard – Trust Observability
- Expand real-time PoI/PoUW analytics, validator health views, and endpoint status so operators can actually see trust behavior evolving.

🔹 #L3 Comparison – Self-Managed vs RaaS
- Run apple-to-apple tests on Testnet-1 (RaaS) vs Testnet-1a (self-managed) to measure performance, reliability, and ops tradeoffs.
(Testnet-1b is internal-only for parallel dev validation.)

🔹 Light #MCP Integration (Nice-to-Have)
- Experiment with MCP-style component assignment so the router starts behaving like an agent execution surface (without full protocol rollout yet).

📌 Docs / Plan
- Phase #2 overview & testing focus: https://docs.cortensor.network/community-and-ecosystem/community-testing/testnet-phase-2#testing-focus
- Phase #2 rewards: https://docs.cortensor.network/community-and-ecosystem/community-testing/testnet-phase-2#rewards

Phase #2 is about turning validation into a redundant, cost-efficient, fully instrumented trust layer - the foundation we need before deeper SLA pricing + agentic integrations land in Phase #3 / #4.

https://x.com/cortensor/status/2012497464256033216

https://farcaster.xyz/cortensor/0x1fb7c7c2

🧪 Testnet Phase #2 – Preview Recap (Starts This Weekend – Jan 17)

We're wrapping prep for Testnet Phase #2, which kicks off this weekend (Jan 17). Most docs + infra are now in semi-final shape, and all environments are wired for the next round of testing:
- Testnet-0 — Arbitrum Sepolia (L2)
- Testnet-1 — COR L3 via RaaS
- Testnet-1a — Self-managed COR L3 (comparison target)
- Testnet-1b — Self-managed COR L3 (internal dev environment)

Phase #2 Focus
🔹 Validator v2/v3 (PoI + PoUW)
Tighten quantitative + qualitative scoring so validators are more deterministic, less “flaky,” and better aligned with the rubric.

🔹 Validator Engine – Ollama + Embeddings
Trial Ollama judge models + multiple embedding variants so validators run well on mixed hardware (CPU/GPU) with stable scores.

🔹 QuantitativeStats / QualitativeStats
Grow calibration datasets + tracking so we can tune thresholds, spot drift, and reduce false positives/negatives across domains.

🔹 Redundancy & Trust Stability
Stress multi-run / multi-miner redundancy and aggregation logic to keep trust stable under partial failures or disagreement.

🔹 Router Surfaces – /delegate & /validate
Exercise agent-facing flows, policy hints, multi-run delegation, and validation-heavy workloads to harden these endpoints.

🔹 Oracle & Internal Service Reliability
Verify router APIs, indexers, and dashboard feeds stay fresh + consistent under sustained load and recovery scenarios.

🔹 Infra Redundancy (RPC / Failover / Backups)
Add and test multiple RPC endpoints, validator backups, and ops playbooks so the network survives infra hiccups gracefully.

🔹 Gas Optimization & Runtime Tuning
Profile cortensord + L2 contracts (SessionQueueValidation, SessionPayment, validator storage) and trim gas + runtime overhead.

🔹 Dashboard – Trust Observability
Expose real-time PoI/PoUW metrics, validator health, and endpoint status so operators can actually see trust behavior.

🔹 L3 Comparison – Self-Managed vs RaaS
Run apple-to-apple tests on Testnet-1 (RaaS) vs Testnet-1a (self-managed) to measure performance, reliability, and ops tradeoffs.

(Testnet-1b is internal-only and will be used for parallel dev validation.)
🔹 Light MCP Integration (Nice-to-Have)
Experiment with MCP-style component assignment so the router starts behaving like an agent execution surface (without full protocol rollout yet).

🔹 Incentives & Pricing
- Rewards: moving toward base reward + level bonus, with role count acting more as a capacity/seniority signal than a separate bonus
- Dynamic pricing: SessionPaymentTable is live across all testnets; fees can factor in SLA, model class, execution depth, validator depth once we flip it on during Phase #2 → Phase #3

🔹 Docs (semi-final, only small tweaks expected)
Phase #2 overview & testing focus: https://docs.cortensor.network/community-and-ecosystem/community-testing/testnet-phase-2#testing-focus
Phase #2 rewards: https://docs.cortensor.network/community-and-ecosystem/community-testing/testnet-phase-2#rewards

Next step: final small adjustments this week, then we’ll share the official Phase #2 start announcement.

https://x.com/cortensor/status/2011029888518832240

https://farcaster.xyz/cortensor/0xb0237460

🗓️ 2025 Year-End Wrap - Cortensor

2025 was the year we proved a simple thesis: decentralized inference can be practical if you treat it like infrastructure - not a science project.

We focused on one thing: build the execution loop, validate it under load, and set the foundation for an agent-ready and developer-ready trust layer - so builders can ship reliable AI apps without a high cost barrier.


Minimal E2E MVP is live

We shipped a minimal but real end-to-end pipeline:
Router → Session Queue → Miners → Validators

Payments: Pay-per-use + Stake-to-Use (still early, still needs tuning).


DevNets → Testnet

We completed 6 DevNet phases, kicked off Testnet, and wrapped Testnet Phase #1.

Load testing reached 150k+ tasks executed, validating feasibility under sustained traffic.


Infra & Scaling Bpgrades

Core improvements landed:
- NodePool / NodePoolUtil refactors + data isolation (future SLA routing)
- Session Queue & SessionQueued scaling for high read-throughput + session-heavy workloads
- Dashboard polish for better NodeOps monitoring + rewards/performance visibility


Models + Validation Surfaces Expanded

We rolled out 64+ models (including embedding stacks for Validator v2/v3).

We iterated on /validate (multiple templates) and introduced /delegate as a thin agent-facing SDK surface.


Hackathons pushed the ecosystem forward

We ran 3 hackathons in 2025.

So far, 15+ apps were qualified/rewarded - not just demos, but real feedback loops that pressure-test the network.


Early Agent Rails: #ERC8004 + #x402 + #MCP (experimental)

We ran rough experiments on #ERC8004, #x402, and #MCP to surface blockers early.

Not fully integrated yet - but these are the rails we’ll harden and fold in during 2026.


What Cortensor is becoming

Cortensor is shaping into an execution + trust layer for web3 agents and agentic apps - while staying practical for everyday developers building AI products.

Agents (and apps) delegate work → Cortensor runs redundant inference + proofs → validation/trust signals return → results become usable, reliable, and affordable, without a high-cost barrier to enter the AI app market.


2026 North Star

Build the shared primitives that both agent economies and developers keep needing:

- Execution that scales (reliable inference under load)
- Validation you can verify (PoI/PoUW trust rails)
- Trust signals you can reuse (for routing, SLAs, and agent workflows)
- Low-friction payments (#x402 + agent-native integrations)

That's the path toward becoming the AWS for agent economies - and a builder-friendly execution layer where anyone can ship agentic apps with confidence.


https://x.com/cortensor/status/2006903828713713710

https://farcaster.xyz/cortensor/0x450ea4dc

🧪 Testnet Phase #1 Recap – Integration, Load, and Engine Expansion

Phase #1 was the first structured Testnet week where we focused on one thing: break the pipeline, harden the pipeline, and prove the system scales under real usage across both Testnet-0 (Arbitrum Sepolia) and Testnet-1 (COR #L3).

🔹 Incentivized Manual Testing (E2E + Payments)
- Daily manual testing by community testers across different models and task types
- End-to-end validation: routing → miner execution → validator checks → results
- Exercised both Pay-to-Use (pay per task) and Stake-to-Use (stake-based access / rate-limits)
- General stress-testing across components to surface edge cases early

🔹 Automated Load Testing – 130K+ Tasks Executed
- Automated task jobs ran continuously during Phase #1
- 130,000+ tasks executed, validating stability and throughput under sustained traffic

🔹 Dashboard UI Refinement
- Multiple rounds of UI polish across session/task views
- Improved navigation and usability during high-volume sessions

🔹 #Ollama Engine Support + 65+ Models Live
- Added #Ollama LLM engine support across Testnet-0 and Testnet-1
- 65+ models are now supported (Q1 target: 100+)
- Current limitation is disk footprint: nodes need many models stored to serve on request
- Next up: network-driven model auto-assignment + model-usage-aware routing
- Also added additional embedding models via Ollama to support Validator v2/v3 tuning, since the current embedding stack shows GPU-specific issues on some setups

🔹 SessionQueue Scaling – Read Throughput & Pagination
- Phase #1 load exposed a read bottleneck at high task volumes
- Added missing foundations: precomputed task stats per session + scalable pagination
- Dashboard + Router can now handle large query/read workloads under single sessions

🔹 Corgent Experimental – /delegate + /validate
- First iterations to validate rough E2E behavior and clarify exact requirements
- Established what needs to be strict vs flexible for agent-facing flows

🔹 Router Experiments: #MCP + #x402 + #ERC8004
- #MCP, #x402 and #ERC8004 components are functional inside Router Node today
- Still require refinement and full integration
- These will be re-iterated during Phase #3–#4 to land as router-native primitives

🔹 Roadmap 2026 Docs Ready (Cortensor Network + Corgent)
- Roadmap documentation is now published and aligned for 2026 planning

Phase #1 proved the pipeline can survive both real usage and sustained automated load. Next phases will build on this foundation toward SLA-driven routing, tighter payment flows, and deeper agentic integrations.

https://x.com/cortensor/status/2005225713801277648

https://farcaster.xyz/cortensor/0xc94e762e

🧪 Testnet Phase #1 Starts Now

The first of four structured Testnet phases is now live.

We're entering a 1-week grace period: node stats and session data have been reset on Testnet-0 (Arbitrum Sepolia) and Testnet-1 ($COR Rollup L3), and this week serves as onboarding + warm-up before full regression.

🔹 Focus – System Integration & Regression
- Validate all node roles and coordination flows end-to-end across apps, routing, miners, and validators
- Stress-test both Pay-to-Use and Stake-to-Use inference flows under live conditions
- Exercise the new NodePool refactor to generate large user-task & validator datasets for future SLA filters and inference-quality tuning

🔹 Incentivized Testing & Bug Bounty
- Node ops: submit random sessions/tasks daily to simulate real usage
- Manual + load testing throughout Phase #1 will be incentivized
- Bugs and edge cases will qualify for bounty rewards
- Detailed rules and amounts will be announced soon in Discord

🔹 Docs / Plan
- Testnet Phase #1: https://docs.cortensor.network/community-and-ecosystem/community-testing/testnet-phase/testnet-phase-1
- Testnet Phases: https://docs.cortensor.network/community-and-ecosystem/community-testing/testnet-phase
- Structured Plan: https://docs.cortensor.network/community-and-ecosystem/community-testing/testnet-phase-wip-draft/wip-draft-testnet-phases-and-focus-areas-structured-plan

Phase #1 is about breaking and hardening the pipeline so later phases can safely introduce SLA-driven routing, payments, and agentic integrations.

https://x.com/cortensor/status/1989954731058630714

https://farcaster.xyz/cortensor/0x9d6ac074

🧪 Phase #6 Recap – Final Dev Phase Before Testnet (Latest)

Phase #6 wraps as one of Cortensor's most complete milestones - the final dev phase before Testnet. Core modules are functional, network task validation is live, and the agentic roadmap through 2026 is defined.

🔹 Validator Expansion & Stabilization
- v2 (PoI) & v3 (PoUW) live on DevNet7, Testnet-0, Testnet-1; end-to-end network task validation confirmed.
- v2 shows hardware-specific variance - fine-tuning continues during Testnet.

🔹 SessionPayment – Stake-to-Use Prototype
- SessionPaymentStaking + SessionPaymentStakingRateLimit powering Stake-to-Use.
- E2E prototype complete; dashboard shows live rate-limit usage and resets.

🔹 Dashboard & Visualization Enhancements
- Session/Task views: staking mode, rate-limit usage, miner participation.
- Validator views: quantitative + qualitative breakdowns; improved network stats.

🔹 Hackathon #2
- Submissions closed with the Phase #6 snapshot; tools and early agentic prototypes inform Testnet challenges.

🔹 Testnet Preparation
- All modules are up to date with DevNet6 and DevNet7.
- Testnet targets identical behavior and quality to DevNet6/DevNet7.

🔹 Testnet Plans & Docs – 4 Phases
- Mid-Nov ’25 → early/mid-Q2 ’26; ~4–6 weeks/phase on Testnet-0 (Arbitrum Sepolia) & Testnet-1 (COR L3).

🔹 Router → Agentic Surface (x402 + ERC-8004)
- High-level brainstorming/design for Router’s shift to an agentic entry surface; forms the backbone for the 3-phase integration through and after Testnet.

🔹 x402 + ERC-8004 (COR Prover)
- 3-phase path: Phase A COR Prover PoC (validation-only in Testnet Phase 4) → Phase B/C Router Agent (inference + validation post-Testnet).
- Aligns Router v2/v3, Session contracts, and external agents.

🔹 COR/ETH on Base (Uniswap)
- Initial COR/ETH pair launched on Base, bridged from Ethereum issuance via native Base Bridge.

🔹 Conclusion
Phase #6 moves Cortensor from prototype to Testnet-ready architecture - unifying validation, staking, payments, privacy groundwork, and agentic planning. Next: launch the 4-phase Testnet and begin the march to agentic AI infrastructure through 2026.

https://x.com/cortensor/status/1984965147337855428

https://farcaster.xyz/cortensor/0xf8e461d4

Cortensor Announcements pinned a video

Welcome to the Official Cortensor Community!

Cortensor is transforming AI inference through our decentralized, community-powered network, making advanced AI tools accessible to everyone.

Important Links:
• Website: https://www.cortensor.network
• Documentation: https://docs.cortensor.network
• X / Twitter: https://x.com/cortensor
• LinkTree: https://linktr.ee/cortensor

🔔 "Tap to verify" button below and join our community group to engage in real-time discussions, get the latest updates, and be part of the movement!

Click below to verify you're human

🧪 Phase #6 Recap – Final Dev Phase Before Testnet

Phase #6 has been one of the most comprehensive milestones yet - the final development phase before the Testnet rollout. With about two weeks remaining, nearly all planned objectives are complete, and the network is operating in one of its most stable states to date.

🔹 Validator Expansion & Stabilization
- Both Validator v2 (PoI) and Validator v3 (PoUW) are fully live and running across DevNet7, Testnet-0, and Testnet-1.
- Node operators are now participating in distributed validation, confirming stability across both GPU and CPU instances.
- Visualizations, heatmaps, and validator leaderboards provide real-time insights into validation performance.

🔹 SessionPayment Integration
- Full contract-to-contract payout and real-time event validation are functional.
- End-to-end payment verification runs continuously across test environments.

🔹 Memory Index & GPU Benchmarks
- Dynamic model loading, switching, and NodePool matching stabilized.
- Network-wide benchmarking for LLaMA 3.x, DeepSeek, and other supported models completed successfully.

🔹 Dashboard & Data Visualization
- Expanded validator stats dashboards and improved real-time visibility for quantitative and qualitative metrics.
- Validator detail views and indexer endpoints provide deeper insights into task validation patterns.

🔹 Hackathon #2 – Active Phase
- Three-Four weeks in, builders continue shipping agentic web apps and decentralized AI tools.
- About two weeks remain until submission close (Nov 2).

🔹 Deferred Items → Testnet Phase
- Privacy-preserving sessions and deep stress testing will continue into the upcoming Testnet phases, when more real-world traffic and user data become available.

Phase #6 has successfully transitioned Cortensor from prototype validation to a multi-network, multi-role system with live validators, payments, and visual analytics.

The final weeks will focus on validation data refinement, final node onboarding, and preparing for the public Testnet launch.

https://x.com/cortensor/status/1979669575395672549

https://farcaster.xyz/cortensor/0x0844831e

🧪 Phase #6 Starts Now

The final development phase before Testnet is live.

We're entering a 1-week grace period. Most nodes are already migrated to DevNet7 with a few still finalizing. Once migration completes, stats will reset and tasks will run across the network to confirm stability.

What's in Phase #6:
🔹 Node Level Rewards
- Base rewards introduced by tier (3 levels)
- Linear level bonus on top

🔹 Memory Index & GPU Benchmarks
- General dynamic model loading, switching, reporting & NodePool matching
- Stress testing all supported models (CPU + GPU: incl. LLaMA 3.x, DeepSeek)

🔹 SessionPayment Integration
- Full contract ↔ contract payout flow
- Real-time event trigger validation

🔹 Regression – Refactored Sessions
- Session & SessionQueue data isolation fully live
- Backward compatibility + regression testing

🔹 Oracles & Canary Testing
- Multi-region shards monitored across CA, UT, OH
- Canary rollout on DevNet7 + Testnet-0/1

🔹 Privacy Prototyping
- Early privacy-preserving sessions & queues prototyping
- Full evaluation deferred to Testnet due to limited user data in DevNet

🔹 Validator Expansion Prototyping
- Extending Validator v1 with quantitative (embedding/distance) + qualitative (LLM scoring) checks
- Prototyping PoI & PoUW modules; full rollout during Testnet

🔹 Hackathon #2
- Launches one week after Phase #6 start
- Focused on real-world apps, integrations, and agent tools
- Double rewards pool compared to Hackathon #1
- Draft/WIP: https://docs.cortensor.network/community-and-ecosystem/hackathon/hackathon-2

🔹 Testnet Stress Testing
- Run existing hackathon apps & demos on Testnet-0/1
- Validate user task flows and cross-env stability

📄 Full details: https://docs.cortensor.network/community-and-ecosystem/community-testing/closed-alpha-testing-phase-6

📍 Network DevNet7 Stats: https://dashboard-devnet7.cortensor.network/stats/network

🌉 L3 Bridge: https://cortensor-testnet-1-32ph80a02h-01acc12814e9cea3.testnets.rollbridge.app

🔎 L3 Explorer: https://testnet1.explorer.cortensor.network

💬 Join the community: https://discord.gg/cortensor

👉 The last dev phase before Testnet – stability, scaling, and trust features all under test.

https://x.com/cortensor/status/1967144644623692212

https://farcaster.xyz/cortensor/0x80c92b9f

🧪 Phase #5 Recap – Scaling Toward Testnet Readiness

Phase #5 pushed Cortensor through its most advanced testing so far, focusing on SLA-driven routing, validator scoring, memory-aware orchestration, and ecosystem stress.

🔹 SLA-Based Node Assignment (v1)
- Session creation now supports SLA tiers (Low / Medium / High)
- NodePool selects based on Node Level with fallback logic
- More refinement done during live testing to improve routing fairness

🔹 Validator v1 + Node Level Runtime
- Validator tracked task success/failure across nodes
- Upgrade/downgrade logic applied after 10+ results
- Improved scoring precision and SLA enforcement under load

🔹 CognitiveLevel – Adaptive Node Categorization
- Miner-side logic categorized nodes by runtime performance
- Validator-side SLA enforcement validated routing decisions
- Early foundation for recursive inference and smarter orchestration

🔹 SessionPayment – Tightly Coupled Flow
- Loosely coupled, event-driven payouts validated and stable in Phase #5
- Tightly coupled Session ↔ SessionPayment integration completed
- First end-to-end payments executed directly between modules
- Further testing planned during Phase #6, including delta calculations for charging extra features

🔹 Dual Testnet Setup
- Testnet-0 (Arbitrum Sepolia): full-stack task pipeline validated
- Testnet-1 (L3 COR Rollup): ~100 miners live, L3 stress tests stable

🔹 User Task Mode Flexibility
- Dedicated Nodes: supported multi-model setups (e.g., LLaMA 3.1, DeepSeek-R1)
- Ephemeral Nodes: longer reservations + model switching tested

🔹 NodeSpec Memory Index
- Rolled out across environments, classifying nodes by system/GPU VRAM
- Enables memory-aware task routing by matching sessions to nodes that can actually load the required model
- Will be fully activated in Phase #6 / DevNet7

🔹 Hackathon #1
- Ran in parallel with Phase #5
- Builders delivered live apps: AI assistants, market/sentiment bots, decentralized oracles, Telegram bots, and more
- Entries showcased how Cortensor infra supports real-world decentralized inference

🔹 UX + Dashboard Updates
- Session creation flow updated with SLA configs
- Node stats, L3 interaction, and memory index visibility added

📊 Stats: https://dashboard-devnet6.cortensor.network/stats/heatmap/rank/table

👉 Phase #5 tightened SLA routing, validated memory-aware orchestration, and advanced SessionPayment into tightly coupled mode - all while fostering ecosystem apps via Hackathon #1. With DevNet7 prepped, we now move into Phase #6: the final development phase before Testnet.

https://x.com/cortensor/status/1962103740699398493

https://farcaster.xyz/cortensor/0xc41e7970

Cortensor Announcements pinned a video