FLOP miner rewards are distributed based on the proportion of inference computing power, Arthur Hayes says allocation should be adjusted based on actual results
Odaily reports: Arthur Hayes posted on X platform that FLOP miners' block rewards are calculated based on the proportion of inference computing power they process. For example, if a single block processes half of the total volume, they receive half of the block reward. He stated that if purchased hardware is suited for a particular AI model but demand for that model is insufficient, miners may be unable to process inference requests and earn rewards; one can refer to popular models on OpenRouter to predict the models needed for agent work and adjust allocation based on actual results.
