Axon Research Bittensor mining, engineered.

01 What we do

Operations first. Research and engineering make it work.

• Mining

  We operate miners

  We run miners on a small set of selected subnets and earn emissions by producing the work their validators reward.

• Research

  We dissect incentive mechanisms

  We take apart how each subnet scores miners, test what actually moves the score, and publish the findings.

• Engineering

  We build reproducible infrastructure

  Versioned, scripted, and measured. Every setup can be rebuilt from scratch and carried to the next subnet.

02 Approach

Consistency comes from method, not capital or luck.

1. P1

   Data over hype

   Decisions come from measurements: validator scores, logs, and controlled experiments. Not from rumours or narratives.

2. P2

   Understand the mechanism first

   Before we deploy on a subnet, we read its code and work out what the incentive mechanism really rewards.

3. P3

   Reproducible infrastructure

   If a result can't be rebuilt from a clean machine, it doesn't count. Infrastructure lives in code, not in someone's head.

4. P4

   Open by default

   We publish part of what we learn so the community can check our work, and so trust is earned rather than claimed.

03 Current focus

Subnets we work on

• SN56

  Gradients

  Decentralized model training: miners compete to fine-tune text and image models on submitted datasets, and validators score the results.

  Open the SN56 Scout: live tournament intelligence for SN56 Gradients
• SN105

  Beam

  Decentralized bandwidth: miners move data across distributed infrastructure, and validators score transfers on measured throughput, latency, and reliability.

04 Research

Published findings

• Coming soon

  Our first public report

  We're preparing our first write-up. Follow @axonresearchxyz on X to see it when it's published.