As reported on November 19, 2026, Pearl’s FP8 upgrade had gone live on its testnet the previous day. The question is whether a change in supported computing formats can help Pearl turn AI inference into useful mining work—or whether miners can still earn PEARL without doing that work.
How Pearl differs from Bitcoin
Pearl is a Bitcoin code fork, but its proposed mining model differs from Bitcoin’s conventional proof of work. Bitcoin rewards miners with newly issued BTC and transaction fees for securing the network; its block subsidy halves roughly every four years. Pearl instead aims to let AI inference providers submit proofs of computations they already perform for customers and earn block rewards through a mechanism it calls Duplex Work.
That aim is not yet a requirement. Bankless analyst David Christopher noted that, under the mechanism described, miners can still receive tokens for blocks that do not represent genuine AI inference. The upgrade does not by itself establish that rewarded work serves real users.
What FP8 changes
Pearl previously supported INT7 for qualifying inference work. According to Christopher’s analysis, providers using other formats would have had to convert their models to participate. Moving the testnet mining standard to FP8 could reduce that obstacle: the analysis says more than 80% of model inference traffic on API aggregation platforms such as OpenRouter uses FP8.
The trade-off is hardware access. INT7 can run on a wider range of older graphics cards, while FP8 typically requires newer GPU architectures. Christopher also cautioned that FP8 support does not prevent meaningless computation from being passed off as useful work; more sophisticated on-chain verification would still be needed.
Adoption and token-incentive questions
The original report described a rise in PEARL over the preceding week and cited listings on Coinbase and Bybit alongside interest in the FP8 testnet. A price move, however, does not demonstrate adoption by AI providers. Christopher identified several tests for Pearl’s model: whether established providers publicly acknowledge mining PEARL and pass token-supported API discounts to customers; whether mining power becomes less concentrated; and what share of blocks actually corresponds to inference requested by real users. He noted that the last measure is particularly difficult to observe externally.
Crypto researcher hitesh.eth offered a more skeptical assessment. He argued that the case for Pearl depends on Bitcoin miners moving toward AI computing because of token incentives, and that incentives sustained by high token inflation may not last. He also alleged coordinated influencer promotion and questioned whether higher token prices would translate into product adoption. Those are his criticisms, not established outcomes. The FP8 testnet change addresses a format barrier, but the reported evidence does not yet resolve the separate questions of useful work, verification or durable demand.
