Some of the links on this site may be affiliate links. ICOBench may receive advertising commissions when users visit recommended platforms via our affiliate links, at no additional cost to you. All recommendations are based on a comprehensive review process. 
In the lastest Solana news, Solana’s mainnet target slot time is expected to move from 400 milliseconds to 350 milliseconds in epoch 1020, following a one-epoch delay after the feature activated at the start of epoch 1019. The effective change is set for epoch 1020.
Solana’s August 6 engineering changelog confirms that the feature gate for lowering slot times from 400ms to 350ms was live on Devnet and Testnet. The changelog also points readers to an upcoming feature-gate schedule, while listing releases and engineering work across the Solana ecosystem.
The target timing is not the same as observed block production, transaction-confirmation latency, or economic finality. The change shortens the target production interval without increasing the compute allowance per second.
EXPLORE: Best Memecoins Presales to Watch in 2026
Solana News: What the 350ms Target Changes
Reducing the target interval from 400ms to 350ms shortens it by 50 milliseconds. The theoretical increase in scheduling frequency is approximately 14.3%, calculated as 400 divided by 350 minus 1. That calculation describes the target cadence only; it is not a forecast for transaction throughput or confirmation speed.
The staged reduction to draft proposal SIMD-0525. Under the report’s description of the proposal, the mainnet feature account was activated in epoch 1019, while the effective 350ms target is delayed until epoch 1020.
🚨we're activating 1st slot time reduction on mainnet
help ensure devs are ready because the transition period is wonky…
1. We have NOT updated some sdk constants like `DEFAULT_MS_PER_SLOT` to reflect new values. Note we will push a release with updated values post activation…
— Brennan Watt (@bw_solana) August 18, 2026
The same report describes a reduction in the example maximum compute budget per block as target slots become shorter. It lists 100 million compute units at a 400ms target and 87.5 million compute units at 350ms. Both figures correspond to a theoretical maximum of roughly 250 million compute units per second: 100 million divided by 0.4 seconds and 87.5 million divided by 0.35 seconds.
That theoretical ceiling is not a transaction-throughput forecast. Actual activity depends on workload and network conditions, and a maximum block-compute limit does not mean every block contains that amount of work.
Solana would continue assigning four consecutive slots to each leader. At a 400ms target, that corresponds to a nominal 1.6-second leader window; at 350ms, it corresponds to 1.4 seconds. The shorter span leaves less time for validators to receive a previous block, replay it, build on it, and land votes before the network moves to the next slot.
EXPLORE: Best Memecoins Presales to Watch in 2026
The Staged Path Toward 200ms
The proposed sequence of target-slot reductions from 400ms to 350ms, then 300ms, 250ms, and 200ms. The Solana news report says Testnet was operating at an effective 200ms target, while Devnet was at 300ms and had activated a 250ms gate that was not yet effective.
The 350ms mainnet change should therefore be viewed as one stage in a br
The first slot time reduction feature is now live on Solana testnet.
SIMD-0525 is active on testnet, reducing slot time from 400ms to 350ms as the first of four feature-gated steps toward 200ms slot time. pic.twitter.com/8XnTu4Byqs
— Solana Developers (@solana_devs) August 5, 2026
oader proposed sequence rather than as confirmation that every later target will take effect on the same schedule. SIMD-0525 remains a draft, and the activation of a feature on a cluster does not make the full 200ms design an accepted final standard.
The August 6 changelog provides additional context on validator-client work. It lists releases for Agave and Firedancer, describes additional conformance work on Agave, and says Mithril is keeping updated on Agave’s conformance suite. The changelog says Mithril is a new validator implementation written in Go that will soon participate in consensus, and that its work with the conformance suite is intended to help it keep up with Agave.
The changelog also describes improvements to Agave’s leader Turbine broadcasting cache and routing-table rebuild. According to the changelog, leaders create blocks from transactions sent by the network, split blocks into shards, and distribute them to validators through Turbine.
