Gonka community portal, created and maintained by community members

Gonka approved proposals to change its model lineup and run a three-month PR campaign with a budget of 75,000 USDT.

LLM tokens this week
22B
DevShard · input + output
at least Pool trading volume
$97k
Uniswap v3 WGNK/USDT 0.3% on Ethereum
Reporting period
22—28
June 22–28, 2026, UTC
LLM tokens per day
LLM tokens per day02.3B4.7B22 Jun28 Jun
Current periodPrevious period
DevShard · input + output
GNK price · daily close, USD
GNK price · daily close, USD$0.271$0.286$0.30222 Jun28 JunIncomplete data · 167/168

GNK trading volume: at least $97.6 thousand

Measured GNK trading volume totaled at least $97.6 thousand. At the start of the period, the price in the tracked pool was $0.3121. No closing price is available, so the weekly price change was not calculated. Volumes cannot be compared with the previous week because the hourly data is incomplete.

Fewer completed requests, more tokens

DevShard statistics show 2,111,349 completed requests, down 12.1% from June 15–21. A total of 18.61 billion input tokens and 3.37 billion output tokens were processed. Their combined volume increased by 7.7%. MiniMax-M2.7 accounted for 66.9% of token volume, followed by Kimi-K2.6 and Qwen3-235B.

Model and PR proposals approved

Proposal #78 passed: it retains MiniMax-M2.7 as the sole model for PoC compute verification and the base model for delegation, removing Qwen and Kimi from the lineup. Proposal #79, to add Kimi K2.6 and GLM 5.2, was then approved. Approval does not mean that both models are already live.

Proposal #77 for a PR campaign passed. The three-month plan includes publications, podcasts, speaking engagements and an industry report. The budget is 75,000 USDT, with a target of at least 35 publications. The budget approval does not confirm that funds have been transferred.

Response quality and context checks

Participants discussed incoherent Kimi responses and switches to Chinese. Views on the causes varied. Some proposed testing long context during validation and publishing errors by host. Opinions on artificial load were divided: some saw it as a way to uncover failures, while others considered it an unnecessary expense for hosts.

Plan published for another attempt to connect Kimi

On June 27, a plan for the next attempt to connect Kimi to PoC at epoch 311 was published in the documentation. It calls for advance announcements of participation and a window for switching nodes. A previously added notice stating that Kimi was active from epoch 309 was removed. A page for submitting vulnerability reports through HackerOne was also added.

← All editions