Pioneering Decentralized AI Inference, Democratizing Universal Access. #AI #DePIN ๐Ÿ”— linktr.ee/cortensor

Pinned Tweet
PSA: There are NO official tokens for Bardiel or Corgent. No #BARDIEL token No #CORGENT token Any token claiming to be โ€œBardielโ€ or โ€œCorgentโ€ is not launched or endorsed by us โ€“ treat it as a scam. Always verify via official @Cortensor channels before interacting with anything on-chain. #Cortensor #Bardiel #Corgent #ScamAlert #AgenticAI
6
51
111
9,638
๐Ÿ› ๏ธ DevLog โ€“ RPC Separation Between Testnet and Mainnet Environments Is Underway A quick infra update. ๐Ÿ”น Current progress - The newer servers for the RPC side are now ready / provisioned - Weโ€™re now working on the next step of separating the RPC environments more cleanly ๐Ÿ”น What this means - The main direction is to separate: - testnet RPC environment - mainnet RPC environment ๐Ÿ”น Why this matters - Cleaner separation helps reduce infra overlap between environments - It also makes the RPC side easier to manage as we move further toward Mainnet Lite and broader mainnet prep - This is also one of the earlier prerequisite items for Mainnet Full infra readiness later on ๐Ÿ”น Current status - This is still WIP for now - but the new server side is ready and the separation work is actively underway #Cortensor #DevLog #RPC #Infra #MainnetLite #MainnetFull
๐Ÿ› ๏ธ DevLog โ€“ New RPC Servers Ordered, Infra Separation Comes Next A quick infra update. ๐Ÿ”น Current progress - Weโ€™ve ordered a new server for the RPC path - It is now getting fulfilled, and weโ€™re waiting on the data center response ๐Ÿ”น Whatโ€™s next - Later this week, weโ€™ll start separating some of the infra across servers - This is also part of the earlier pre-prep work for Mainnet Lite ๐Ÿ”น Why this matters - The main goal is to separate the RPC environments more cleanly - In practice, that means moving toward clearer separation between: - testnet RPC infra - mainnet RPC infra ๐Ÿ”น Current takeaway - new RPC capacity is on the way - infra separation comes next - and this is part of the broader Mainnet Lite pre-prep path #Cortensor #DevLog #RPC #Infra #MainnetLite
1
4
11
215
๐Ÿ› ๏ธ DevLog โ€“ Cortensor L3 Is Ending Phase #4 in a Much Stronger Place A quick recap on where the Cortensor L3 path stands as we move toward the end of Testnet Phase #4. ๐Ÿ”น Current scale so far - The L3 path has now processed: - over 8M total blocks - over 16.5M total transactions That gives a much better picture of where the network stands now versus where it started. ๐Ÿ”น Why this matters - This phase was not only about keeping the chain running - It was also about exercising the broader infra path under real usage, repeated task flow, and product-facing traffic on top ๐Ÿ”น Recovery / infra ops side - We also had to work through a real recovery scenario during this phase - That was painful, but it was also useful - It helped prove more about: - recoverability - rebuild / reindex flow - L3 infra ops readiness - how the stack behaves under edge cases instead of only normal paths ๐Ÿ”น Current takeaway - As Phase #4 comes to an end, the Cortensor L3 path is in a much more robust place - not just because of raw block / tx count - but because it has now gone through both sustained usage and real infra recovery scenarios too #Cortensor #DevLog #L3 #Testnet1a #AIInfra #Arbitrum
โฐ Reminder โ€“ Testnet Phase #4 Wraps This Weekend A quick reminder on the near-term timeline. ๐Ÿ”น Current phase - Testnet Phase #4 is expected to finish / wrap up this weekend ๐Ÿ”น What comes with that - prize processing / follow-up comes after the wrap-up - then we move into the planning stage for Mainnet Lite prep ๐Ÿ”น What happens next - More detailed Mainnet Lite prep plans will be shared after the planning period in the weeks following the Phase #4 wrap-up ๐Ÿ”น Current takeaway - Phase #4 is almost done - wrap-up and prizes come next - then Mainnet Lite planning begins #Cortensor #DevLog #Testnet #Phase4 #MainnetLite
1
11
18
259
โฐ Reminder โ€“ Testnet Phase #4 Wraps This Weekend A quick reminder on the near-term timeline. ๐Ÿ”น Current phase - Testnet Phase #4 is expected to finish / wrap up this weekend ๐Ÿ”น What comes with that - prize processing / follow-up comes after the wrap-up - then we move into the planning stage for Mainnet Lite prep ๐Ÿ”น What happens next - More detailed Mainnet Lite prep plans will be shared after the planning period in the weeks following the Phase #4 wrap-up ๐Ÿ”น Current takeaway - Phase #4 is almost done - wrap-up and prizes come next - then Mainnet Lite planning begins #Cortensor #DevLog #Testnet #Phase4 #MainnetLite
โฐ Reminder โ€“ Testnet Phase #4 Wraps in About a Week A quick reminder on the broader timeline. ๐Ÿ”น Current phase - Testnet Phase #4 is expected to wrap in about a week ๐Ÿ”น What comes after that - Weโ€™ll finish wrapping up Phase #4 fully - then shift more attention toward the planning and prep work for Mainnet Lite, as mentioned earlier ๐Ÿ”น Current takeaway - about one more week for the current phase - then Phase #4 wrap-up - then more focused Mainnet Lite planning and prep from there #Cortensor #DevLog #Testnet #Phase4 #MainnetLite
7
14
429
Mainnet Lite comes first for a reason. A more practical and controlled L2 path on @Arbitrum, built for early hosted access, developer APIs, and real mainnet validation before the fuller stack rolls out. It is also the earlier demo / playground-style path where the system can be used, tested, and refined under real mainnet conditions before broader expansion. Q3 2026 has started, and Mainnet Lite is the first mainnet step we are building toward in this phase. #Cortensor #MainnetLite #Mainnet #Arbitrum
Mainnet Lite is the controlled first step for bringing the Cortensor stack onto real mainnet conditions. The point is not to do everything at once. It is to start with a simpler, more productized L2 path first at @Arbitrum, validate the stack under real conditions, and then build outward from there. Q3 2026. #Cortensor #MainnetLite #Mainnet #Arbitrum
11
22
418
๐Ÿ› ๏ธ DevLog โ€“ Portal Stress Tests Resume With Larger Prompts and Token Sizes A quick follow-up on the current Portal testing path. ๐Ÿ”น Current status - Weโ€™ve resumed the current tests - We are also increasing prompt size and overall token size as part of this next round ๐Ÿ”น What this means - The testing is no longer only around smaller request patterns - Weโ€™re now pushing heavier prompt / completion usage through the Portal path as well ๐Ÿ”น Current focus - larger prompt sizes - larger token output - continued multi-model routing checks - continued admin / ops visibility while the load profile changes ๐Ÿ”น Whatโ€™s next - Weโ€™ll keep running this through the rest of the week - and use it to keep checking routing, latency, quota behavior, and token-pressure patterns under heavier request shapes #Cortensor #DevLog #Portal #StressTest #APIGateway #AIInfra
๐Ÿ› ๏ธ DevLog โ€“ Portal Admin / Ops Visibility Keeps Getting More Useful A quick follow-up on the Portal admin / ops side. ๐Ÿ”น Current progress - Weโ€™ve kept refining the admin / ops visibility surface so it is easier to understand what is happening across the Portal path - That includes broader improvements around metrics, hotspot visibility, and operational drilldowns ๐Ÿ”น What improved - We added more triage-oriented views to help identify noisier or weaker paths faster - That includes triage around: - pool routes - model lanes - API Gateway instances - mapped sessions underneath ๐Ÿ”น Why this matters - As traffic grows, it becomes harder to rely only on raw tables or broad summary metrics - These triage views make it easier to spot: - bad nodes - bad sessions - weaker model paths - noisier gateway instances ๐Ÿ”น Current takeaway - Portal admin / ops is no longer only about broad observability - it is also becoming more useful for actual triage and follow-up when something starts degrading #Cortensor #DevLog #Portal #Observability #APIGateway #AIInfra
1
9
22
365
๐Ÿ› ๏ธ DevLog โ€“ Portal Admin / Ops Visibility Keeps Getting More Useful A quick follow-up on the Portal admin / ops side. ๐Ÿ”น Current progress - Weโ€™ve kept refining the admin / ops visibility surface so it is easier to understand what is happening across the Portal path - That includes broader improvements around metrics, hotspot visibility, and operational drilldowns ๐Ÿ”น What improved - We added more triage-oriented views to help identify noisier or weaker paths faster - That includes triage around: - pool routes - model lanes - API Gateway instances - mapped sessions underneath ๐Ÿ”น Why this matters - As traffic grows, it becomes harder to rely only on raw tables or broad summary metrics - These triage views make it easier to spot: - bad nodes - bad sessions - weaker model paths - noisier gateway instances ๐Ÿ”น Current takeaway - Portal admin / ops is no longer only about broad observability - it is also becoming more useful for actual triage and follow-up when something starts degrading #Cortensor #DevLog #Portal #Observability #APIGateway #AIInfra
๐Ÿ› ๏ธ DevLog โ€“ Portal Stress Tests Resumed, Monitoring Continues A quick follow-up on the current Portal testing path. ๐Ÿ”น Current status - Portal stress tests have resumed - So far, things look good overall ๐Ÿ”น Current issue - One model / node path still needs some attention from node ops - That has already been flagged and shared on our side ๐Ÿ”น Visibility - The public Portal stats surface is reflecting the current state - The internal admin / ops pages are reflecting it as well ๐Ÿ”น Whatโ€™s next - Weโ€™ll keep these tests running through this week - and continue monitoring while we wrap up the current Testnet Phase #4 ๐Ÿ”น Current takeaway - stress tests are back in motion - overall path looks fine so far - one node/model path needs follow-up - monitoring continues across both public stats and internal ops surfaces #Cortensor #DevLog #Portal #StressTest #Observability #AIInfra
1
8
18
589
๐Ÿ› ๏ธ DevLog โ€“ New RPC Servers Ordered, Infra Separation Comes Next A quick infra update. ๐Ÿ”น Current progress - Weโ€™ve ordered a new server for the RPC path - It is now getting fulfilled, and weโ€™re waiting on the data center response ๐Ÿ”น Whatโ€™s next - Later this week, weโ€™ll start separating some of the infra across servers - This is also part of the earlier pre-prep work for Mainnet Lite ๐Ÿ”น Why this matters - The main goal is to separate the RPC environments more cleanly - In practice, that means moving toward clearer separation between: - testnet RPC infra - mainnet RPC infra ๐Ÿ”น Current takeaway - new RPC capacity is on the way - infra separation comes next - and this is part of the broader Mainnet Lite pre-prep path #Cortensor #DevLog #RPC #Infra #MainnetLite
๐Ÿ—“๏ธ Weekly Focus โ€“ Phase #4 Wrap-Up & Mainnet Lite Preparation This week is mainly about wrapping up Testnet Phase #4. Most Phase #4 work is now complete, so the focus shifts toward final validation before entering the Mainnet Lite prep period, which is expected to last around ~6 weeks. ๐Ÿ”น Phase #4 โ€“ Wrap-Up & Final Validation - Continue monitoring routing, miners, validators, dashboards, indexers, and L3 stats through the final week. - Run another overall validation pass to ensure Phase #4 closes with a clean baseline. ๐Ÿ”น Portal V1 โ€“ Product Refinement - Continue polishing Portal V1 across auth, API keys, usage visibility, request logs, and overall UX. - Tighten the hosted product flow before shifting attention toward Mainnet Lite deployment. ๐Ÿ”น Portal V1 โ€“ Public Stats Refinement - Continue refining the new public Portal stats page built on live production data. - Improve caching, observability, and overall UX while expanding useful public metrics. ๐Ÿ”น Portal V1 โ€“ Final E2E Validation - With Testnet-1a nearly recovered, run another round of Portal E2E testing. - Validate the full hosted path across Gateway โ†’ router pools โ†’ router nodes under real conditions. ๐Ÿ”น Mainnet Lite โ€“ Final Baseline Checks - Re-check the remaining Mainnet Lite baseline items before entering the prep period. - Confirm infra, RPC, contracts, dashboard, indexer, oracle, and application wiring are all ready. ๐Ÿ”น Mainnet Lite โ€“ Prep Planning - Continue preparing the Mainnet Lite rollout plan, infrastructure, and application setup. - Mainnet Lite remains the controlled first mainnet step before the broader Mainnet Full/L3 direction. ๐Ÿ”น PyClaw โ€“ Dev Path Progress - Continue PyClaw iteration toward the first public development release. - Keep improving workflow, tooling, and side packages before the initial public snapshot. This week is about closing Phase #4 cleanly, validating Portal V1 under real conditions, and preparing everything needed before transitioning into the Mainnet Lite preparation period. #Cortensor #Testnet #Phase4 #AIInfra #DePIN #Portal #PyClaw #MainnetLite #L3
1
6
14
463
โฐ Reminder โ€“ Testnet Phase #4 Wraps in About a Week A quick reminder on the broader timeline. ๐Ÿ”น Current phase - Testnet Phase #4 is expected to wrap in about a week ๐Ÿ”น What comes after that - Weโ€™ll finish wrapping up Phase #4 fully - then shift more attention toward the planning and prep work for Mainnet Lite, as mentioned earlier ๐Ÿ”น Current takeaway - about one more week for the current phase - then Phase #4 wrap-up - then more focused Mainnet Lite planning and prep from there #Cortensor #DevLog #Testnet #Phase4 #MainnetLite
๐Ÿ—“๏ธ Weekly Focus โ€“ Phase #4 Wrap-Up & Mainnet Lite Preparation This week is mainly about wrapping up Testnet Phase #4. Most Phase #4 work is now complete, so the focus shifts toward final validation before entering the Mainnet Lite prep period, which is expected to last around ~6 weeks. ๐Ÿ”น Phase #4 โ€“ Wrap-Up & Final Validation - Continue monitoring routing, miners, validators, dashboards, indexers, and L3 stats through the final week. - Run another overall validation pass to ensure Phase #4 closes with a clean baseline. ๐Ÿ”น Portal V1 โ€“ Product Refinement - Continue polishing Portal V1 across auth, API keys, usage visibility, request logs, and overall UX. - Tighten the hosted product flow before shifting attention toward Mainnet Lite deployment. ๐Ÿ”น Portal V1 โ€“ Public Stats Refinement - Continue refining the new public Portal stats page built on live production data. - Improve caching, observability, and overall UX while expanding useful public metrics. ๐Ÿ”น Portal V1 โ€“ Final E2E Validation - With Testnet-1a nearly recovered, run another round of Portal E2E testing. - Validate the full hosted path across Gateway โ†’ router pools โ†’ router nodes under real conditions. ๐Ÿ”น Mainnet Lite โ€“ Final Baseline Checks - Re-check the remaining Mainnet Lite baseline items before entering the prep period. - Confirm infra, RPC, contracts, dashboard, indexer, oracle, and application wiring are all ready. ๐Ÿ”น Mainnet Lite โ€“ Prep Planning - Continue preparing the Mainnet Lite rollout plan, infrastructure, and application setup. - Mainnet Lite remains the controlled first mainnet step before the broader Mainnet Full/L3 direction. ๐Ÿ”น PyClaw โ€“ Dev Path Progress - Continue PyClaw iteration toward the first public development release. - Keep improving workflow, tooling, and side packages before the initial public snapshot. This week is about closing Phase #4 cleanly, validating Portal V1 under real conditions, and preparing everything needed before transitioning into the Mainnet Lite preparation period. #Cortensor #Testnet #Phase4 #AIInfra #DePIN #Portal #PyClaw #MainnetLite #L3
1
8
15
494
๐Ÿ”Ž Recap: What We Mean by โ€œAgent Cloudโ€ A quick recap on one of the longer-term directions behind Cortensor. ๐Ÿ”น The simple framing Agent cloud does not just mean agents calling an API. It means agents can: - work locally - use hosted model access - delegate tasks outward - call remote subagents - validate results before acting - return outcomes into one connected workflow ๐Ÿ”น Why it matters As agents get more capable, they should not have to do everything in one place. Some work stays local. Some work uses hosted inference. Some work gets delegated outward through remote trust and execution layers. That is the shift from โ€œone agent, one runtimeโ€ toward a broader execution fabric. ๐Ÿ”น How the stack connects - PyClaw = local-first agent workflows - Portal = hosted model access - Corgent / Bardiel = remote delegation, validation, and trusted execution - Cortensor Network = routing, validation, and execution underneath So the stack is not just one app or one endpoint. It is a broader execution environment that agents can build on. ๐Ÿ”น Why /delegate matters /delegate is what starts making this real. Instead of every agent doing everything itself, work can be handed off into another execution surface and brought back into the same flow. That is why the โ€œAWS for agentsโ€ framing exists: - remote execution - delegated work - capability-aware routing - trust and validation - reusable infrastructure underneath And that direction also lines up with the broader โ€œintelligent delegationโ€ idea now being discussed more openly: delegation is not just calling another model, but deciding who should do what, under what constraints, with what oversight, and how the result should be verified. ๐Ÿ”น Current takeaway Agent cloud, in the Cortensor sense, means agents can move work across local, hosted, and remote delegated paths on top of one broader execution layer. That is the direction connecting PyClaw, Portal, Corgent, Bardiel, and /delegate over time. #Cortensor #AgentCloud #AgenticAI #Corgent #Bardiel #PyClaw
1
9
17
341
๐Ÿ› ๏ธ DevLog โ€“ Portal Stress Tests Resumed, Monitoring Continues A quick follow-up on the current Portal testing path. ๐Ÿ”น Current status - Portal stress tests have resumed - So far, things look good overall ๐Ÿ”น Current issue - One model / node path still needs some attention from node ops - That has already been flagged and shared on our side ๐Ÿ”น Visibility - The public Portal stats surface is reflecting the current state - The internal admin / ops pages are reflecting it as well ๐Ÿ”น Whatโ€™s next - Weโ€™ll keep these tests running through this week - and continue monitoring while we wrap up the current Testnet Phase #4 ๐Ÿ”น Current takeaway - stress tests are back in motion - overall path looks fine so far - one node/model path needs follow-up - monitoring continues across both public stats and internal ops surfaces #Cortensor #DevLog #Portal #StressTest #Observability #AIInfra
๐Ÿ› ๏ธ DevLog โ€“ Testnet1a Is Back to Normal, Portal Stress Tests Resume Later Today A quick follow-up on the recent testnet1a recovery. ๐Ÿ”น Current status - L3 / testnet1a is back to normal - All router endpoints are back to normal as well ๐Ÿ”น Why this matters - This was a good recovery exercise for us on the infra / ops side - It gives more confidence that, with the current dataset and recovery path, we can rebuild and recover from this kind of disaster event more cleanly than before ๐Ÿ”น Whatโ€™s next - Weโ€™ll keep monitoring things further today - If everything continues to look stable, weโ€™ll resume the Portal stress tests later today ๐Ÿ”น Public stats - Public Portal stats is also reflecting again: - portal.cortensor.network/staโ€ฆ ๐Ÿ”น Current takeaway - testnet1a recovery looks successful - router endpoints are back - monitoring continues today - Portal stress testing should resume later today if things keep holding #Cortensor #DevLog #Testnet1a #Portal #Infra #L3
9
15
564
๐Ÿ› ๏ธ DevLog โ€“ Testnet1a Is Back to Normal, Portal Stress Tests Resume Later Today A quick follow-up on the recent testnet1a recovery. ๐Ÿ”น Current status - L3 / testnet1a is back to normal - All router endpoints are back to normal as well ๐Ÿ”น Why this matters - This was a good recovery exercise for us on the infra / ops side - It gives more confidence that, with the current dataset and recovery path, we can rebuild and recover from this kind of disaster event more cleanly than before ๐Ÿ”น Whatโ€™s next - Weโ€™ll keep monitoring things further today - If everything continues to look stable, weโ€™ll resume the Portal stress tests later today ๐Ÿ”น Public stats - Public Portal stats is also reflecting again: - portal.cortensor.network/staโ€ฆ ๐Ÿ”น Current takeaway - testnet1a recovery looks successful - router endpoints are back - monitoring continues today - Portal stress testing should resume later today if things keep holding #Cortensor #DevLog #Testnet1a #Portal #Infra #L3
๐Ÿ› ๏ธ DevLog โ€“ L2 / L3 RPC Upgrade Is Planned for Later Today A quick infra note. ๐Ÿ”น Current update - The L2 / L3 RPC path will be upgraded later today - This is tied to a mandatory Arbitrum-side upgrade ๐Ÿ”น Environments affected - Testnet0 - Testnet1a - Mainnet Lite ๐Ÿ”น What this means - We expect these environments to be affected during the upgrade window - This is mainly part of keeping the RPC/base infra aligned with the required upstream changes ๐Ÿ”น Current direction - Weโ€™ll work through the upgrade later today - then continue monitoring the affected environments after that #Cortensor #DevLog #RPC #Arbitrum #Infra
1
5
14
504
๐Ÿ› ๏ธ DevLog โ€“ L2 / L3 RPC Upgrade Is Planned for Later Today A quick infra note. ๐Ÿ”น Current update - The L2 / L3 RPC path will be upgraded later today - This is tied to a mandatory Arbitrum-side upgrade ๐Ÿ”น Environments affected - Testnet0 - Testnet1a - Mainnet Lite ๐Ÿ”น What this means - We expect these environments to be affected during the upgrade window - This is mainly part of keeping the RPC/base infra aligned with the required upstream changes ๐Ÿ”น Current direction - Weโ€™ll work through the upgrade later today - then continue monitoring the affected environments after that #Cortensor #DevLog #RPC #Arbitrum #Infra
๐Ÿ› ๏ธ DevLog โ€“ Testnet1a Recovery: L2 Batch + DA Rebuild Looks Good So Far A quick follow-up on the testnet1a recovery path. ๐Ÿ”น Current progress - It now looks like the full blocks have been synced / rebuilt / reindexed successfully using the L2 batch path and the available DA dataset - That gives us a much better signal that the recovery path is holding up ๐Ÿ”น Whatโ€™s next - Weโ€™ll run more E2E checks later today through the Portal path - That should help confirm the broader product/API flow on top of the recovered chain state ๐Ÿ”น Why this matters - This is useful not only for restoring testnet1a - It is also a good exercise for us on the infra / ops side around: - L3 recovery flow - edge cases - data recoverability - rebuild / reindex procedures under failure conditions ๐Ÿ”น Current takeaway - testnet1a recovery looks much healthier now - Portal-side E2E checks come next - and the recovery process itself has been a useful infra / ops exercise for the longer-term L3 path #Cortensor #DevLog #Testnet1a #Infra #L3 #Portal
1
7
12
518
๐Ÿ› ๏ธ DevLog โ€“ Testnet1a Recovery: L2 Batch + DA Rebuild Looks Good So Far A quick follow-up on the testnet1a recovery path. ๐Ÿ”น Current progress - It now looks like the full blocks have been synced / rebuilt / reindexed successfully using the L2 batch path and the available DA dataset - That gives us a much better signal that the recovery path is holding up ๐Ÿ”น Whatโ€™s next - Weโ€™ll run more E2E checks later today through the Portal path - That should help confirm the broader product/API flow on top of the recovered chain state ๐Ÿ”น Why this matters - This is useful not only for restoring testnet1a - It is also a good exercise for us on the infra / ops side around: - L3 recovery flow - edge cases - data recoverability - rebuild / reindex procedures under failure conditions ๐Ÿ”น Current takeaway - testnet1a recovery looks much healthier now - Portal-side E2E checks come next - and the recovery process itself has been a useful infra / ops exercise for the longer-term L3 path #Cortensor #DevLog #Testnet1a #Infra #L3 #Portal
๐Ÿ› ๏ธ DevLog โ€“ Testnet1a Recovery Is Still in Progress A quick update on the current testnet1a recovery path. ๐Ÿ”น Current status - The sequencer is synced again - But there still appears to be some validation correction / cleanup happening on top of that - We have a few recovery steps lined up and will keep working through them today ๐Ÿ”น What this means - The recovery is moving forward - But it is not yet at the point where we want to call it fully clean until those remaining checks are done ๐Ÿ”น One lesson from this process - One clear takeaway from the testnet1a recovery process is that we should have additional sequencer backup coverage as well - Even though we had already planned to run multiple sequencers in production / mainnet, we were more relaxed on testnet because it is โ€œjust testnetโ€ and the infra cost multiplies quickly ๐Ÿ”น What changes from here - That backup / redundancy thinking is now more serious, even for environments where we previously accepted more risk for cost reasons - This is already under progress as part of the follow-up thinking from this incident ๐Ÿ”น Current takeaway - testnet1a recovery is still ongoing - the sequencer is synced, but a few correction / validation checks remain - and this incident is reinforcing the need for stronger sequencer backup planning going forward #Cortensor #DevLog #Testnet1a #Infra #Sequencer
2
10
18
556
๐Ÿ› ๏ธ DevLog โ€“ Portal Stats UI/UX Is Getting More Refined A quick follow-up on the Portal public stats side. ๐Ÿ”น What improved - We made broader UI/UX refinements across the Portal stats surface - That includes: - dark theme support - better mobile behavior - more responsive layouts across screen sizes ๐Ÿ”น Caching follow-up - We also adjusted cache timing by interval so stats queries do not run more often than needed - Current direction: - 14-day view = 2h cache - 7-day view = 1h cache ๐Ÿ”น Why this matters - As more data accumulates, these stats queries and aggregations become heavier - So we want to be more careful about query frequency and rely more on cache where it makes sense ๐Ÿ”น Current takeaway - Portal stats is getting more polished on the UI side - and the data side is also being tuned more carefully for performance as the stats surface grows ๐Ÿ”น Whatโ€™s next - Weโ€™ll keep monitoring this more carefully as we do further stress tests and the amount of stats data keeps increasing #Cortensor #DevLog #Portal #Stats #Observability #AIInfra
๐Ÿ› ๏ธ DevLog โ€“ Portal Public Stats Prototype Is Now Live in Rough Form A quick follow-up on the public Portal stats direction. ๐Ÿ”น Current progress - Weโ€™ve now pushed the rough shape / initial prototype live - It is reading from actual data rather than only static placeholders - Current ranges: - portal.cortensor.network/staโ€ฆ - portal.cortensor.network/staโ€ฆ ๐Ÿ”น What this means - This is the first early public-facing stats surface for Portal - It gives a simpler public view into service status, request outcomes, latency, token throughput, request size, and recent status timeline ๐Ÿ”น Performance follow-up - We also added a shared cache layer - That should help pages load faster - and reduce unnecessary load on query / summary / aggregation paths underneath ๐Ÿ”น Current takeaway - Public Portal stats is no longer only an internal concept - the first rough prototype is now live - and weโ€™ll keep refining it further from here #Cortensor #DevLog #Portal #Observability #Stats #AIInfra
3
12
18
492
๐Ÿ”Ž Recap: How PyClaw Could Connect to Corgent / Bardiel as an Agent Cloud A quick recap on the longer-term direction for how these pieces can connect. ๐Ÿ”น PyClaw side - PyClaw is closer to the local-first agent and app experience - It gives developers a more familiar surface for building and running workflows locally ๐Ÿ”น Corgent / Bardiel side - Corgent is the more infra-native trust / execution surface - Bardiel is the more product-facing trust + execution layer on top - Together, they are closer to the remote delegation / validation / execution side of the Cortensor stack ๐Ÿ”น Where the dots connect - The important future direction is not only โ€œPyClaw uses LLMs through Portalโ€ - It is also that PyClaw agents could call Corgent / Bardiel endpoints for remote delegation and remote subagent execution So the rough picture becomes: - local agent starts in PyClaw - some work stays local - some work gets delegated outward - remote subagents run through Corgent / Bardiel-style execution paths - trust / validation / routing happens underneath on Cortensor ๐Ÿ”น Why this matters - That starts to look much more like an agent cloud model - local agents do not need to do everything themselves - they can hand off tasks to remote delegated agents when needed - and the remote side is not just generic API calling, but a trust-aware execution layer ๐Ÿ”น What that means in practice - PyClaw = local agent/workflow surface - Portal = hosted model access layer - Corgent / Bardiel = remote delegation and trusted execution layer That combination is what starts connecting the dots into something bigger than a single app or a single endpoint. ๐Ÿ”น Current takeaway The missing piece is not the idea itself, but the integration path over time. If that path gets built out, PyClaw would not just be a local agent tool. It could become a front-end surface for agents that can spawn, delegate to, and coordinate remote subagents through the Cortensor stack. That is where the โ€œagent cloudโ€ direction starts to become real. #Cortensor #PyClaw #Corgent #Bardiel #AgenticAI #AIInfra
7
14
300
๐Ÿ—“๏ธ Weekly Focus โ€“ Phase #4 Wrap-Up & Mainnet Lite Preparation This week is mainly about wrapping up Testnet Phase #4. Most Phase #4 work is now complete, so the focus shifts toward final validation before entering the Mainnet Lite prep period, which is expected to last around ~6 weeks. ๐Ÿ”น Phase #4 โ€“ Wrap-Up & Final Validation - Continue monitoring routing, miners, validators, dashboards, indexers, and L3 stats through the final week. - Run another overall validation pass to ensure Phase #4 closes with a clean baseline. ๐Ÿ”น Portal V1 โ€“ Product Refinement - Continue polishing Portal V1 across auth, API keys, usage visibility, request logs, and overall UX. - Tighten the hosted product flow before shifting attention toward Mainnet Lite deployment. ๐Ÿ”น Portal V1 โ€“ Public Stats Refinement - Continue refining the new public Portal stats page built on live production data. - Improve caching, observability, and overall UX while expanding useful public metrics. ๐Ÿ”น Portal V1 โ€“ Final E2E Validation - With Testnet-1a nearly recovered, run another round of Portal E2E testing. - Validate the full hosted path across Gateway โ†’ router pools โ†’ router nodes under real conditions. ๐Ÿ”น Mainnet Lite โ€“ Final Baseline Checks - Re-check the remaining Mainnet Lite baseline items before entering the prep period. - Confirm infra, RPC, contracts, dashboard, indexer, oracle, and application wiring are all ready. ๐Ÿ”น Mainnet Lite โ€“ Prep Planning - Continue preparing the Mainnet Lite rollout plan, infrastructure, and application setup. - Mainnet Lite remains the controlled first mainnet step before the broader Mainnet Full/L3 direction. ๐Ÿ”น PyClaw โ€“ Dev Path Progress - Continue PyClaw iteration toward the first public development release. - Keep improving workflow, tooling, and side packages before the initial public snapshot. This week is about closing Phase #4 cleanly, validating Portal V1 under real conditions, and preparing everything needed before transitioning into the Mainnet Lite preparation period. #Cortensor #Testnet #Phase4 #AIInfra #DePIN #Portal #PyClaw #MainnetLite #L3
๐Ÿ—“๏ธ Weekly Recap โ€“ Phase #4 Portal Maturity & Mainnet Lite Baseline Complete This week Portal V1 continued to mature into a real hosted product, while Mainnet Lite closed out nearly all remaining baseline validation work. ๐Ÿ”น Phase #4 โ€“ Monitoring, Support & Stats - Continued monitoring across routing, miners, validators, dashboards, indexers, and L3 stats. - Phase #4 remained stable while Portal V1 became the primary engineering focus. ๐Ÿ”น Portal V1 โ€“ Product Refinement - Portal continued evolving with improvements across auth, API keys, request visibility, usage reporting, and overall UI/UX. - The hosted product flow now feels much closer to a practical MVP than an early prototype. ๐Ÿ”น Portal V1 โ€“ Token Metering & Multi-Model Routing - Token-based usage metering moved into rough implementation across the Gateway, Portal, and admin surfaces. - Multi-model routing expanded into live testing across oss-20b, Gemma4, Qwen, and oss-120b. ๐Ÿ”น Portal V1 โ€“ Gateway, Stress Tests & Operations - Portal stress tests expanded into heavier workloads and multi-hour runs. - API Gateway scaled to three instances while admin/ops metrics and token-level observability continued improving. ๐Ÿ”น Mainnet Lite โ€“ Baseline Validation Complete - Dedicated-node user-task, ephemeral-node network-task, and ephemeral-node user-task โ†’ node-pool flows were all validated successfully. - With payment distribution revalidated, the Mainnet Lite baseline checklist is now largely complete. ๐Ÿ”น Mainnet Lite โ€“ Infrastructure & Operations - Additional ephemeral nodes were added for cleaner validation, while internal RPC infrastructure remained stable. - A sequencer recovery exercise on Testnet-1a also highlighted the need for stronger backup/redundancy planning going forward. ๐Ÿ”น Payment Staking โ€“ Regression & Hardening - Additional regression and hardening checks looked good on Testnet-1a. - Follow-up validation will continue, but no major issues have surfaced. ๐Ÿ”น PyClaw โ€“ Dev Path Progress - Prepared the first public development version (early 0.0.1-style snapshot) for release. - Iteration continues toward a more usable framework during Q3 while remaining open for public feedback. A productive Phase #4 week overall - Portal V1 made another significant step toward a real hosted inference product, while Mainnet Lite baseline validation is no longer the primary blocker and attention is steadily shifting toward product refinement and operational maturity. #Cortensor #Testnet #Phase4 #AIInfra #DePIN #Portal #PyClaw #MainnetLite #L3
1
5
13
892
๐Ÿ› ๏ธ DevLog โ€“ Current Phase Ends Soon, Then Mainnet Lite Prep Begins A quick reminder on the broader timeline and what comes next. ๐Ÿ”น Current phase timing - The current Testnet Phase #4 is expected to finish in about a week - After that, weโ€™ll start wrapping things up and move into the Mainnet Lite prep period ๐Ÿ”น What Mainnet Lite prep means - This is not the same thing as immediate full rollout - The Mainnet Lite prep period is more about: - re-checking infra baseline checks - re-checking the remaining checklist items - setting up and configuring apps on the Mainnet Lite @Arbitrum L2 path - wiring those apps into router endpoints and the broader hosted path correctly ๐Ÿ”น Expected prep window - The Mainnet Lite prep period will likely last around 4โ€“6 weeks - During that period, the apps/products will be set up again in a way similar to how we prepared and iterated through testnet1a ๐Ÿ”น Mainnet Lite expectation - Mainnet Lite should be thought of as the more controlled first mainnet step - It is closer to a hosted / demonstration-style rollout and practical product path, not the final full network shape all at once ๐Ÿ”น In parallel - During Q3, we will also keep looking into Mainnet Full / L3 infra and baseline work in parallel - So Mainnet Lite prep does not mean Mainnet Full thinking stops - it just means Lite comes first as the earlier rollout path ๐Ÿ”น Current takeaway - Testnet Phase #4 is close to finishing - Mainnet Lite prep comes next for around 4โ€“6 weeks - Lite is the earlier controlled mainnet path - and Mainnet Full / L3 baseline work will also continue during Q3 #Cortensor #DevLog #MainnetLite #MainnetFull #Arbitrum #AIInfra
๐Ÿงช 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 โ€” data โ†’ quality โ†’ routing, reliability under load, and Mainnet Lite readiness. ๐Ÿ”น E2E scope lock (Phase #3 โ†’ Phase #4) Finalize the end-to-end test plan + order + pass/fail metrics (throughput/latency, recovery, drift, data integrity/offchain correctness) ๐Ÿ”น Deeper stress testing (Data โ†’ Quality Path) Longer, higher-volume runs that hit real failure modes (miner churn, partial outages, RPC degradation, queue spikes) ๐Ÿ”น Privacy + offchain storage stress tests (optional / heavier) Heavier testing where enabled: encryption paths, offchain correctness, and edge cases (timeouts, stale keys, partial writes) ๐Ÿ”น SLA / quality signals โ€“ deeper validation (#3 signals) Stress SLA signals under churn/load (drift detection, reliability weighting vs uptime, disagreement/arbitration behavior) ๐Ÿ”น Portal V1 (design โ†’ early implementation) + product boundary Portal becomes a real product layer: client โ†’ Portal backend โ†’ router pool โ†’ sessions Scope: API keys, usage visibility, managed router pools, auth/rate limits, billing hooks + metering ๐Ÿ”น Portal focus (gateway โ†’ router pool path) Key issuance/consumption path is proven; Phase #4 drills into the hosted execution shape from Portal Gateway into managed router pools ๐Ÿ”น PyClaw (early dev release track) Move from design into early dev release: landing page + docs + initial runtime scaffolding aligned to router/session primitives ๐Ÿ”น 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) ๐Ÿ”น Core Docs Testnet Phase #4: docs.cortensor.network/commuโ€ฆ #Cortensor #Testnet #Phase4 #DevUpdate #Portal #PyClaw #InferenceQuality #AIInfra #DePIN #L3
1
5
14
369
Monthly Node Rewards โ€“ June 2026 We've now sent out the June 2026 ethereum:0x8e0eef788350f40255d86dfe8d91ec0ad3a4547f node rewards to 24 nodes as part of our monthly node reward program. ๐Ÿ”น Rewards sent to 24 nodes total ๐Ÿ”น Includes dedicated nodes + ephemeral nodes ๐Ÿ”น Recognizing uptime, stability, and contribution ๐Ÿ’ฌ Join our Discord to set up a node or get involved: discord.gg/cortensor Thanks to all node operators pushing decentralized AI forward - itโ€™s ethereum:0x8e0eef788350f40255d86dfe8d91ec0ad3a4547f or nothing. #Cortensor #AI #DePIN #NodeRewards #COR
Monthly Node Rewards โ€“ May 2026 We've now sent out the May 2026 $COR node rewards to 24 nodes as part of our monthly node reward program. ๐Ÿ”น Rewards sent to 24 nodes total ๐Ÿ”น Includes dedicated nodes + ephemeral nodes ๐Ÿ”น Recognizing uptime, stability, and contribution ๐Ÿ’ฌ Join our Discord to set up a node or get involved: discord.gg/cortensor Thanks to all node operators pushing decentralized AI forward - itโ€™s $COR or nothing. #Cortensor #AI #DePIN #NodeRewards #COR
6
15
309
๐Ÿ› ๏ธ DevLog โ€“ Portal Public Stats Prototype Is Now Live in Rough Form A quick follow-up on the public Portal stats direction. ๐Ÿ”น Current progress - Weโ€™ve now pushed the rough shape / initial prototype live - It is reading from actual data rather than only static placeholders - Current ranges: - portal.cortensor.network/staโ€ฆ - portal.cortensor.network/staโ€ฆ ๐Ÿ”น What this means - This is the first early public-facing stats surface for Portal - It gives a simpler public view into service status, request outcomes, latency, token throughput, request size, and recent status timeline ๐Ÿ”น Performance follow-up - We also added a shared cache layer - That should help pages load faster - and reduce unnecessary load on query / summary / aggregation paths underneath ๐Ÿ”น Current takeaway - Public Portal stats is no longer only an internal concept - the first rough prototype is now live - and weโ€™ll keep refining it further from here #Cortensor #DevLog #Portal #Observability #Stats #AIInfra
๐Ÿ› ๏ธ DevLog โ€“ Public-Facing Portal Stats Is Moving Into Early Prototype A quick follow-up on the public Portal stats direction we mentioned earlier. ๐Ÿ”น What it is - We now have a rough prototype for a public-facing stats surface on Portal - The goal is to give people a simpler, higher-level view of how the hosted path is behaving without exposing all the deeper internal admin / ops detail ๐Ÿ”น What it shows - aggregate service status - request outcomes and completion rate - latency trends - token throughput - average request size - a recent status timeline with more customer-visible service notes ๐Ÿ”น Why this matters - Until now, most of the richer observability has lived inside internal admin / ops panels - This public stats surface is meant to turn part of that into something easier to follow from the outside ๐Ÿ”น Current status - This is still an early prototype - But we expect to release this early prototype later today on both: - Prod Portal - Dev0 Portal ๐Ÿ”น Current takeaway - public-facing Portal stats is now moving from internal concept into actual product surface - the first version will be simple and early - then weโ€™ll expand it further from there over time #Cortensor #DevLog #Portal #Observability #Stats #AIInfra
1
10
22
2,049
๐Ÿ› ๏ธ DevLog โ€“ Public-Facing Portal Stats Is Moving Into Early Prototype A quick follow-up on the public Portal stats direction we mentioned earlier. ๐Ÿ”น What it is - We now have a rough prototype for a public-facing stats surface on Portal - The goal is to give people a simpler, higher-level view of how the hosted path is behaving without exposing all the deeper internal admin / ops detail ๐Ÿ”น What it shows - aggregate service status - request outcomes and completion rate - latency trends - token throughput - average request size - a recent status timeline with more customer-visible service notes ๐Ÿ”น Why this matters - Until now, most of the richer observability has lived inside internal admin / ops panels - This public stats surface is meant to turn part of that into something easier to follow from the outside ๐Ÿ”น Current status - This is still an early prototype - But we expect to release this early prototype later today on both: - Prod Portal - Dev0 Portal ๐Ÿ”น Current takeaway - public-facing Portal stats is now moving from internal concept into actual product surface - the first version will be simple and early - then weโ€™ll expand it further from there over time #Cortensor #DevLog #Portal #Observability #Stats #AIInfra
2
11
21
833