Solana's rent cut is one blog post deep, and most of it lands before the last gate

Start with the gate schedule, because it is not evenly spaced. The third step, 5,080 down to 2,575, removes 2,505 lamports per byte on its own, about 40 percent of the whole reduction ^[20]. By the end of the fourth step the rate is 1,322, which is 19 percent of where it began ^[21]. Most of the cut is in force well before the final gate opens, so anyone holding off until step five to re-price a bulk mint is quoting a stale number in the meantime.

What survives the cut is the mechanism underneath it. Accounts are permitted to hold less and none are pushed down to the new minimum ^[8]. Nothing closes an account either: sell a token to zero and its empty ATA stays open, holding the deposit ^[11]. An account funded before activation therefore still returns about 0.002 SOL when closed, while one funded after full activation returns about 0.0002 ^[10], the same factor of ten as the rate change itself ^[22].

That asymmetry is where the writeup's framing starts doing work the arithmetic does not. Its adoption evidence is one sentence: tens of millions of empty accounts already closed, with far more still sitting open ^[14]. No denominator is given and no date. It then confirms that closing an account is a standard instruction the holder's own wallet signs, with the refund sent to an address of their choosing ^[17]. Both statements are probably accurate. They are also the two things you would put in front of somebody before offering to sweep their wallet for them, and there is no named tool here to test that reading against.

The provenance deserves the same flat treatment. This is a single dev.to post, republished from xroot.dev, whose author says the figures were verified against the proposal text rather than the headlines ^[26]. That is the right instinct, and it is still self-attested and unreplicated. The claim that Agave 4.2 mainnet activations began rolling out in late August 2026 rests on the same one document ^[5].

Operationally the two halves of this have different evidentiary needs. Closing accounts you already control does not depend on when any gate opens, and the largest single pools are not in traders' token accounts: an interrupted solana program deploy strands a buffer holding rent for the entire binary, often 1 to 5 SOL per failed attempt ^[15]. That reclaim is available today at the old rate whatever the schedule does. Re-pricing is the opposite case. It depends entirely on dates one writeup asserts, which argues for quoting new-account costs from the rate you can observe on chain rather than the rate the table promises.