For 10 hours and 35 minutes on 2026-07-19, this address did almost one thing: it ran Pyth price-feed maintenance workflows. Of 9,077 recorded transactions in the complete frozen history, 9,071 asserted update-price, two asserted native-asset transfers, and four submitted no commands and never began execution. None reported an execution failure.
The repetition is striking. Transactions arrived with a median gap of 0 milliseconds, a mean gap of 4.2 seconds, and a 90th-percentile gap of 20.5 seconds. That zero median reflects bursts: many transactions share the same millisecond. Activity touched 586 minutes, the busiest being 13:09 UTC with 36 transactions, and the longest idle period was about 160.5 seconds. This cadence is consistent with scheduled or automated execution, but the evidence does not establish who operated the address.
One workflow, two variants
The dominant exact composition appeared 8,470 times, or 93.31% of all recorded transactions: sui/update-price/verified-accumulator-three-update/v1. The smaller variant, sui/update-price/verified-accumulator-batch/v1, appeared 601 times, or 6.62%. Both used the same two package revisions, 0x04e20ddf36af412a4096f9014f4a565af9e812db9a05cc40254846cf6ed0ad91 and 0x5306f64e312b581766351c07af79c72fcb1cd25147157fdc2f8ad76de9a3fb6a.
A representative transaction, DRMD3X6G, verified a cross-chain message, built authenticated Pyth price records, split fee funding, ran three price-related calls, and cleaned up temporary record material. The shorter batch example, H3iqemQE, followed the same verification-first structure but ran two price-related calls. These workflows explain how signed data was prepared for feed maintenance; they do not establish that a stored feed became newer, current, or correct.
The latest detailed stream is a separate, capped view of 500 transactions. Every one of those 500 was an update-price transaction: 469 used the three-update variant and 31 used the batch variant. Nothing in that recent stream changes the address’s basic character.
What moved
Only SUI movements are recorded. Across the gas accounting for the complete history, 0.58643248 SUI came in and 8.581839152 SUI went out, for a net gas change of −7.995406672 SUI. The sender-boundary balance record shows 0.58643248 SUI in and 8.58186543 SUI out, or −7.99543295 SUI. These are overlapping accounting views, not amounts to add together.
The same address owned the gas in all 9,077 transactions; no third-party-paid transactions appear here. The data show expenditure on feed-maintenance activity, not a priced economic return, and they do not establish profit or loss.
The interruptions
The four no-command submissions form two tight clusters. Two occurred at 14:29:38 and 14:29:41 UTC, followed seconds later by the native transfer 3ibyqHxr at 14:29:44. A second pair appeared at 20:29:54 and 20:29:56, followed by another transfer at 20:29:59. The timing suggests the empty submissions may be related to the transfers, but the evidence does not establish why.
3ibyqHxr is also the largest gas charge in the address’s history: 0.296509880 SUI. Its recorded shape split native value and passed the resulting objects to a transfer step. The record does not establish the recipient’s identity or whether custody ultimately changed.
This address is best read as a narrowly focused, high-cadence price-feed maintenance identity with two small operational deviations. For a concrete starting point, inspect DRMD3X6G; compare it with the batch variant H3iqemQE; then examine the interruption around 3ibyqHxr and the empty submission 4Ltn7Eb8.