Gonka community portal, created and maintained by community members

Gonka approved proposals for positioning, brand design, and promotion, while MiniMax-M2.7 led by token volume in DevShard statistics.

LLM tokens this week
16.7B
DevShard · input + output
Reporting period
1—7
June 1–7, 2026, UTC
LLM tokens per day
LLM tokens per day02.5B5B1 Jun7 Jun
Current periodPrevious period
DevShard · input + output

Completed requests rose by 89.3%

Completed requests rose to 1,693,011 from 894,161 for May 25–31. The system processed 13.47 billion input tokens and 3.21 billion output tokens. MiniMax-M2.7 accounted for 44.5% of total token volume, ahead of Kimi-K2.6 and Qwen3-235B. Qwen led the previous week.

Promotion budgets approved; deposit increase vote did not go through

Proposals #60–65 passed: 25,000 USDT for positioning, 20,000 USDT for brand guidelines, 50,000 USDT for an influencer pilot, 10,000 USDT for a website, 28,000 GNK for analytics, and 100,000 GNK for digital strategy. Resubmitted proposal #68 for video production at a cost of 70,000 USD also passed. Approval does not confirm that funds have been transferred or work completed.

Proposal #67 to compensate Kimi operators with 946,509.93 GNK passed. Proposal #59 for an advance payment for training videos was rejected. The vote on proposal #69 to raise the minimum deposit to 500 GNK did not go through. Proposal #70, requesting 246,000 GNK for GNK Racers, had no result by the end of the period.

Discussion focused on promotion and product readiness

Views on promotion spending were divided. Some participants suggested testing service providers with small budgets and measurable criteria first. Others objected to driving traffic to an unstable product. Feedback on inference performance also varied: some reported improvements, while others reported errors and unanswered requests when connecting through brokers.

Memory requirements increased; DevShard update under review

The documentation raised the minimum total GPU memory for an MLNode container from 40 to 320 GB and updated the list of supported GPUs. It is unclear whether the new requirements apply to hosts already operating. PR #1289, a separate update to DevShard, the inference runtime, is open for review. It does not affect the main chain; voting is expected to begin after review, and the rollout has not yet been confirmed.

← All editions