Gonka expands its multi-model compute network with MiniMax-M2.7

Gonka has activated MiniMax-M2.7 as the third governance-approved model participating in its multi-model Proof of Compute system. The activation follows the v0.2.13 upgrade and expands the network beyond the Qwen3–235B and Kimi K2.6 model groups.

Following the completion of its bootstrap phase, the model became active at chain epoch 278, officially joining the network’s compute validation framework.

From a single-model network to a multi-model architecture

For much of its early development, Gonka’s compute layer revolved around a single primary model. That changed with the introduction of multi-model Proof of Compute in v0.2.12 and the addition of Kimi K2.6 as the network’s second governance-approved model.

The addition of MiniMax-M2.7 continues that evolution. With three active model groups now participating in Proof of Compute, Gonka now supports Qwen3–235B, Kimi K2.6, and MiniMax-M2.7 within a single validation framework.

Multi-model Proof of Compute allows different AI models to participate in the network simultaneously rather than concentrating validation around a single model.

What changes with MiniMax-M2.7

MiniMax-M2.7 is a large language model developed by MiniMax, with a focus on reasoning, coding, and agent-oriented use cases. The model supports long-context inference and belongs to a newer generation of reasoning-focused AI systems that place a strong emphasis on autonomous task execution and tool use.

Compared with earlier model groups already active in the network, MiniMax-M2.7 introduces another model profile into Gonka’s Proof of Compute framework, with its own validation and weighting parameters.

Following a bootstrap period for infrastructure preparation and participation setup, MiniMax-M2.7 is now fully active within the network’s Proof of Compute system.

Why it matters

The activation of MiniMax-M2.7 marks the latest milestone in Gonka’s transition from a single-model Proof of Compute system to a multi-model compute network.

With three governance-approved models now active, the network’s Proof of Compute system is no longer centered around a single model family. Instead, operators can choose how to participate across multiple model groups, each with its own validation and weighting parameters.

This shift is increasingly reflected at the protocol level. For every governance-approved model, hosts must explicitly choose whether to join directly, delegate support, or opt out altogether. As additional models are introduced, these decisions are expected to play a growing role in how compute resources are distributed across the network.

Disclosure: Contributors to this blog may participate in the Gonka ecosystem and may hold project tokens. Nothing in this publication constitutes financial or investment advice.