MetaFlux 0.8.6 — Rounds commit four times faster, and BTC and ETH return to 100x

This release is live. MetaFlux 0.8.6 swapped in at block height 8,250,000 on chain 114514. The headline change is not a feature — it is the rate at which the chain agrees. Rounds now commit about four times faster, and the network is under a 72-hour stress observation at the new rate.

Rounds commit about four times faster — The chain previously committed a round every 160 ms, or 6.25 rounds per second. Measured twice over separate 60-second windows immediately after this swap, it commits a round every 40 ms, about 24.7 rounds per second. Your order reaches a committed block sooner by the same factor. Nothing about matching, margin or settlement changed to achieve this: the validators propose on a shorter timer, and the consensus rules are untouched. The figure is one measurement at one release, not a fixed property of the chain.

The round rate is not a constant — A four-fold rate increase raises write volume by the same factor, so disk growth, snapshot cadence and the round rate itself were watched for 72 hours after this swap. Treat the figure above as one measurement at one release, not a promise: it has moved between releases before and since. Integrators who poll on a fixed interval need no change; integrators who assume a block interval in their own code must read it from the chain instead.

Blocks arrive faster, but a block is not a confirmation — A faster round does not change what finality means. Wait for the same confirmation you waited for before. What you gain is that the confirmation arrives sooner, not that you need less of it.

BTC and ETH return to 100x — Both markets carry a maximum leverage of 100 again, with a maintenance margin of 0.5% on the first tier. The maintenance figure is what makes the leverage reachable: admission caps leverage from the maintenance requirement and the taker fee, so the previous 3% requirement held the effective ceiling near 29 whatever the published maximum said. Funding caps on both markets are unchanged — the leverage vote deliberately pinned them to their live values rather than let them move as a side effect.

A settlement defect closed at the activation height — When one order filled against several counterparties in the same batch, and the fill flipped the position through zero, the settlement leg could recover the wrong pre-fill size. The account was credited value that no counterparty paid. The fix takes effect at the activation height and not before, because a committed block must replay exactly as it was executed. Balances written before that height are unchanged; the accounting is correct from it onward.

One correction to an earlier reading — For markets that carry an explicit margin ladder, the maintenance ratio published on the market row does not match the ratio the engine applies. The engine resolves the ladder, and the ladder is correct: BTC and ETH require 0.5%. The single figure printed beside the market overstates it. No position is liquidated on the wrong number — the engine never reads that field for a laddered market — but do not size risk from it. The read is corrected in the next release.