As reported on November 19, 2026, Pearl’s FP8 upgrade had gone live on testnet the previous day. The Bitcoin code fork aims to let AI inference providers earn PEARL block rewards by submitting proofs of computations they already perform for customers. The testnet launch is a step toward that design, not evidence that inference providers have adopted it at scale.
What FP8 changes
Bankless analyst David Christopher’s account describes Pearl’s proposed mechanism as Duplex Work: a provider performs an AI inference task and submits proof of that work for mining rewards. Under the earlier mechanism, eligible work was limited to INT7, requiring providers using other precisions to convert their models before participating. The FP8 upgrade changes which inference work can qualify.
The account says more than 80% of model-inference traffic on aggregators such as OpenRouter runs in FP8. It does not provide an underlying dataset or measurement date for that figure, so it should be treated as a reported claim rather than an independently established measure of Pearl’s addressable workload. The change may reduce conversion friction for FP8 users, but it also raises a hardware trade-off: the account says INT7 can run on a wider range of older GPUs, while FP8 generally requires newer GPU architectures.

The unresolved verification problem
Pearl does not currently require miners to perform genuine AI inference, according to the account. Miners can still earn rewards by computing work with no useful inference output. FP8 compatibility therefore does not, by itself, establish that rewarded blocks correspond to customer requests. Distinguishing useful inference from such work remains a separate verification challenge.
Christopher proposes three tests for the coming months: whether established AI providers publicly acknowledge mining PEARL and pass token subsidies through as lower API prices; whether mining power becomes less concentrated; and what share of blocks can be shown to represent inference for real users. The last measure is also the hardest to observe externally. No results for those tests are supplied.
A competing economic critique
Crypto researcher hitesh.eth argues that Pearl’s market narrative depends on Bitcoin miners moving toward AI computation in response to token incentives. He questions whether rewards supported by token inflation could sustain that shift or produce lasting product adoption. He also alleges coordinated influencer promotion around the upgrade. These are his assessments and allegations, not findings established by the testnet launch.
The original report described PEARL as rising over the preceding week despite a weak broader crypto market and cited Coinbase and Bybit listings as sources of liquidity. It supplied no price series, inflation schedule, mining-concentration data or evidence of provider adoption. Those gaps leave the central research question open: whether Pearl can reward verifiable, customer-driven inference rather than merely make additional computation eligible for block rewards.