What's Coming in Gonka v0.2.11

The next Gonka update is shaping up to be one of the biggest since MainNet launch. Here's what the team has been working on.

Faster AI Processing

Every AI request on Gonka involves starting a job, routing it to a GPU validator, and recording the result on-chain. Each step used to take 10-50ms. The team introduced epoch group caching, precomputed statistical calculations (SPRT), and moved developer stats tracking off the critical path. The core inference loop is now substantially faster.

Unified Permission System

The CertiK audit revealed inconsistent permission checks across the codebase — some message types had strict checks, others had none. The team built a single permission framework so every on-chain action now passes through the same security gate. No more gaps.

Zero-Participation Reward Fix

A subtle bug let validators with PoC weight but zero inference assignments collect full epoch rewards without doing any work. The fix ensures nodes must actually contribute compute to get paid.

Stronger Validator Trust Chain

Two BLS cryptography fixes strengthen how validators verify each other. A self-validation fallback was removed — validators must now always be checked by the previous epoch's group. Additionally, DKG consensus now uses slot-weighted voting instead of simple headcount, preventing small validators from outvoting larger stakeholders.

Continuous Proof of Compute

Instead of checking validators only at specific checkpoints, the network now monitors them around the clock. A critical serialization bug was also fixed that would have silently broken this feature in production.

High Availability Groundwork

The team began building infrastructure for running backup nodes — load-balanced full nodes, validator failover via tmKMS, and database sync for API nodes. This foundational work will make the network more resilient in future versions.