Revenue Search: Inside Bittensor Subnet Session with Wouter & Egill from Zeus: Subnet 18
43 snips
Sep 3, 2025 Egill and Wouter from Zeus discuss their innovative weather forecasting API built on BitTensor. They explain how miners compete to provide accurate weather predictions, with up to 38% lower error rates compared to competitors. The duo reveals their focus on energy and commodity markets, as well as the exciting growth plans ahead, including rapid expansion of forecasting variables. Plus, they share insights into their revenue model and plans for a referral program to boost sales. Get ready for a game-changing approach to weather forecasting!
AI Snips
Chapters
Transcript
Episode notes
Miners Produce Superior Forecasts
- Zeus uses BitTensor miners to forecast weather and claims forecasts outperform existing APIs by a large margin.
- They benchmark against open datasets and report up to ~30–40% lower error on temperature forecasts.
Benchmarking Against ERA5 Gold Standard
- Zeus benchmarks predictions against the ERA5 combined ground-truth dataset used by national services.
- They claim miners are 30–40% closer to actual temperature values than competing APIs on average.
Use Graded Scores, Not Winner-Take-All
- Score miners continuously with a graded 0–1 score so many can serve traffic instead of winner-take-all payouts.
- This encourages load balancing and lets smaller miners participate without needing massive resources.