Fren Pet: the network cost of a Bonk

A completed, direct two-transaction Bonk averaged 0.000002004183 ETH ($0.005363), equivalent to 2,004.183 Gwei in total network fees. The gas-weighted L2 execution price was 0.00616689 Gwei per gas. This result covers 3,475 matched Bonk Commit + Bonk Reveal pairs at Base contract 0x0e22B5f3E11944578b37ED04F5312Dfc246f443C. It is not the fee for just one transaction. Computed results, every pair and its transaction hashes.

Observation window: 24 September 2026 03:00:00 UTC through 25 September 2026 03:00:00 UTC, end-exclusive. This is the latest complete, half-hour-aligned 24 hours at the start of research (03:16 UTC on 25 September), deliberately omitting the unfinished interval after 03:00. It is a frozen historical snapshot, not a rolling “now minus 24 hours” value. Blocks 51,714,727–51,757,926, inclusive; timestamps of the adjacent blocks establish both boundaries. Query manifest, boundary block responses.

Results

Included action	Transactions / Bonks	Mean network fee, ETH	Mean network fee, USD	Mean total fee, Gwei	Mean gas consumed
Commit in matched pairs	3,475	0.000001092598	$0.002924	1,092.598	171,848.94
Reveal in matched pairs	3,475	0.000000911585	$0.002439	911.585	148,078.92
One Bonk: Commit + Reveal	3,475	0.000002004183	$0.005363	2,004.183	319,927.86

Gas is a quantity of execution work; Gwei is a denomination of ETH. To avoid conflating them, the table gives the total fee denominated in Gwei, while the headline separately gives Gwei per gas. One ETH = 1 billion Gwei. Neither gas limits nor maximum fee caps are treated as amounts paid. The median Bonk fee was 0.000001973964 ETH ($0.005282); the empirical 95th percentile was 0.000002249279 ETH ($0.006019). Receipt-derived transaction data.

USD convention: all ETH costs are translated at $2,675.835/ETH, the Coinbase ETH/USD spot endpoint, retrieved 2026-09-25T03:20:10.546747+00:00. This is a reporting-time valuation, not what ETH was worth at each transaction's timestamp, nor a claim about dollars actually paid. The exact response is saved locally.

Thirty-minute heatmap

Open the standalone heatmap · Exact 48-bin values in CSV.

Each cell is the arithmetic mean combined Commit + Reveal fee of pairs whose Reveal occurred in that half-hour. Both transactions must fall inside the overall 24-hour window; a Commit may lie in an earlier cell. Intervals are left-inclusive and right-exclusive. USD uses the same fixed quote as the overall average. Darker means more expensive, and n is the number of pairs. Empty intervals would be marked “No observations,” never zero. The overall average weights each Bonk equally; it is not an unweighted average of cell means.

Evidence and method

1. Enumerate activity. Downloaded the Blockscout transaction history for the fixed block range, yielding 8,572 unique direct transactions. The initial full-window response omitted two calls; a fresh full-window response and two separate half-window queries agreed on exactly the same hash set. Both missing receipts were added before computing results, yielding one additional completed pair. The initial response is retained as evidence/transactions-initial.json. Also fetched all 43,656 logs emitted by the contract with eth_getLogs from Base's public RPC in 22 contiguous block ranges. Raw transaction history; individual log responses and queries are in evidence/logs-*.json.
2. Identify the requested actions. Filtered direct calls to the specified address by calldata selectors 0x3935a788 (Bonk Commit) and 0xa4333d91 (Bonk Reveal). Historical decoded examples establish the Commit signature and Reveal signature; those old transactions are not used in the fee sample. Downloaded a receipt for every one of the 7,130 matching current-window calls. Receipt status distinguishes success from reversion. Logs alone cannot supply fees or reveal reverted calls, so transaction inputs and receipts are necessary supplements.
3. Infer and validate pairs. In block/transaction order, matched each successful Reveal to the preceding unmatched successful Commit with the same sender and calldata attacker ID. All 3,475 successful Reveals matched; one successful Commit remained unmatched. Every successful direct call's contract event was reconciled with the independent log download, and all matched Reveals had no intervening Commit event for that attacker anywhere in the contract logs, including wrapped calls. Commit event payloads matched the first four calldata arguments. Reveal validation accepts both observed outcome topics; a successful transaction does not necessarily mean a winning game outcome. Full topic hashes and checks are in the offline analysis. Pairing is an explicit inference from ordered state activity, not a contract-provided unique Bonk transaction-pair ID or a cryptographic verification of the reveal preimage.
4. Compute actual fees. For each receipt, fee_wei = gasUsed × effectiveGasPrice + l1Fee. Sum the two fees per pair, then divide their total by 3,475. ETH = wei / 10^18; total Gwei = wei / 10^9; USD = ETH × the recorded quote. Gas-weighted execution price = total L2 execution fees / total gas used / 10^9. Base documents L2 execution and L1 data/security fees. Three saved explorer cross-checks agree exactly with this formula. No additional fee component is added; no separate operator fee appears in these receipts, and this matches the checked explorer totals. The L1 component accounts for 1.558% of the matched-pair fees.

What a player can take from this

Observed: the cheapest average interval was 2026-09-24 22:30:00 UTC, at $0.005019 per Bonk (113 pairs). The most expensive was 2026-09-24 08:00:00 UTC, at $0.010374 (33 pairs). Their difference is $0.005355 per Bonk, or 106.7% above the cheaper interval. This comparison describes this sample only; gas consumption, game state and player mix can also change a cell's average. It does not establish a recurring cheapest time of day.

Observed: 175 Commit calls and 4 Reveal calls reverted: 2.51% of direct Bonk-related calls. They still consumed 0.000082717863 ETH ($0.221339) in aggregate. These costs are excluded from the successful-pair headline. Spreading those failed-call costs across this window's completed pairs would add $0.000064 per Bonk, but this is a population-level overhead estimate, not an attribution of retries to individual pairs. One unmatched successful Commit is also excluded. All successes and failures.

Inference for budgeting: plan for two transaction fees and some possible failed-call overhead. Of the matched pairs, 3,472 revealed within one minute; the maximum observed delay was 184.5 minutes. That describes behavior, not a safe reveal deadline. This analysis did not establish the game's timing rules, and it does not recommend delaying a Reveal to chase a cheaper cell.

Scope, uncertainty and unanswered questions

• Scope matters: the full log scan contains 18,953 Commit events and 18,952 events with the two observed Reveal-outcome topics. Most therefore occur outside the direct two-transaction sample. The headline answers the requested direct Bonk Commit + Bonk Reveal flow, not the average across every automated, batched or wrapped Bonk in the game. Allocating one wrapper transaction's fee among multiple actions would require a separate attribution model; no such all-game average is claimed.
• Coverage and provider limits: transaction enumeration comes from Blockscout, with split-query consistency checks; independent RPC logs corroborate successful selected actions. A full scan of every block's transactions was not performed, so a provider omission of a reverted transaction cannot be independently ruled out. Saved responses are attributable provider observations, not independently verified block proofs. Blocks were observed after inclusion, not accompanied by a finalized-L1 proof.
• Excluded costs: token transfers, game stakes/payments, approvals, funding or bridge transactions, and wallet/service charges outside these receipts. The estimate measures network fees, not the total economic cost or expected return of playing.
• Unanswered: historical transaction-time USD spending; fees allocated per Bonk inside wrappers; causes of individual reversions; whether this one day's timing pattern persists. No confidence interval or causal timing claim is inferred from the 48 descriptive means.

Reproduce and inspect

From the repository root, run python3 scripts/analyze.py and python3 scripts/write_report.py with Python 3; no network or third-party packages are needed. The first script rebuilds the CSVs, JSON summary and SVG from delivered evidence and asserts receipt coverage, fees, event consistency, pairing, time boundaries and weighted-bin reconciliation. Local check result. These checks were performed locally and are not an independent audit. README describes files and limits.