Gonka Processed 3.5 Trillion AI Tokens in October, Up From 1 Trillion

Gonka, a decentralized AI compute network, processed 3.5 trillion AI tokens in October, a more than threefold increase from the roughly 1 trillion tokens recorded in September, according to figures released by the project. The monthly total translates into an average of approximately 113 billion tokens per day.

The throughput growth coincided with two structural developments: the activation of governance proposal No. 115 and the completion of legal registration for the Gonka Foundation as an independent nonprofit.

What did proposal No. 115 change?

The community approved proposal No. 115 during the same period, activating dynamic model coefficients on the network. For operators and model deployers, the change matters because coefficients of this kind typically govern how work is priced or weighted across different models on a compute marketplace. Put simply, a network handling 113 billion tokens per day needs a mechanism that adjusts model-level parameters as usage scales, rather than relying on fixed values set when volume was a fraction of current levels.

The scale is worth grounding. Processing 3.5 trillion tokens in a single month means the network handled the equivalent of millions of book-length texts — a frame that practitioners tracking decentralized inference capacity will recognize as a meaningful benchmark against centralized alternatives.

The foundation formalizes the organizational layer

The Gonka Foundation has completed legal registration as an independent nonprofit. Its stated remit covers three functions:

• supporting protocol development;
• providing legal representation;
• coordinating partnerships.

Community authorization and budget arrangements for the foundation are advancing through the project's governance processes, meaning the entity's funding and mandate remain subject to token-holder approval rather than executive discretion. For partners evaluating integration, an incorporated foundation offers a counterparty with legal standing — a prerequisite for institutional agreements that informal developer collectives often cannot sign.

Hardware ambitions and upcoming appearances

Gonka co-founder Daniil Liberman said the team is participating in the development of a new chip. He provided no specifications or mass-production timeline, so the initiative remains at an early, undisclosed stage. Involvement at the silicon layer would nonetheless signal intent to move up the stack from network coordination toward physical compute supply, should the effort reach production.

Daniil and David Liberman, the project's co-founders, are scheduled to speak at Blockchain Life in Dubai on December 1–2, where additional detail on both the foundation's roadmap and the chip initiative could surface.

What does this change for practitioners?

The combination of rising throughput, a governance-activated pricing mechanism and a legally incorporated foundation addresses the three questions any serious deployer asks of a compute network: is there demand, is the economics layer adaptable, and is there an accountable entity behind the protocol. October's numbers answer the first; proposal No. 115 begins to answer the second; the foundation registration answers the third in legal terms.

The near-term checkpoint to watch is the completion of the foundation's budget authorization through governance — the step that converts an incorporated shell into a funded operational body — alongside any disclosure on chip specifications expected to follow the Libermans' December appearances in Dubai.