For the first time since launch, Solana has lowered its target slot time from 400 to 350 milliseconds on mainnet. Effective August 21 at the start of epoch 1020, this change, driven by proposal SIMD-0525, opens the path to a four-step roadmap targeting the symbolic 200-millisecond mark.
🔑 Key Takeaways
- Target slot time lowered from 400 to 350 ms, effective from epoch 1020
- Trillium measures a weighted average of 365.4 ms across epoch 1021
- Missed slot rate drops sharply: 0.077% vs 0.438% in epoch 1015
- Per-block compute cap adjusted to 87.5M units (from 100M since July)
- Four successive 50 ms steps scheduled to reach 200 ms by end of cycle
A historic first revision since launch
Until now, the 400-millisecond parameter defined at mainnet launch had never been revised. The change was triggered by the activation of a feature gate (a software lock enabled at a specific slot) at slot 440,208,000, the first slot of epoch 1019, in line with the one-epoch offset mechanism specified in proposal SIMD-0525. This approach gives the change one full cycle to settle before taking effect, allowing operators time to update their infrastructure.
Jacob Creech, VP of Technology at the Solana Foundation, confirmed the transition on X, stating: « We are entering a new 350-millisecond era. » The Foundation then published a follow-up reporting an average slot time near 360 milliseconds shortly after activation, down from the initial 400 ms protocol parameter.
Consensus mechanism and confirmation thresholds
Solana relies on a hybrid consensus combining proof of history (a cryptographic clock that timestamps every event in a verifiable order) and proof of stake (validator selection weighted by staked tokens). Proof of history enables validators to process transactions in parallel by establishing a verifiable order before execution.
Two commitment levels govern transaction finality: confirmed once at least 66% of stake has voted for the block, and finalized once 31 additional blocks have been built on top. At 400 ms, finalization occurred in roughly 13 seconds, equivalent to 32 consecutive slots.
Telemetry data: a stable transition
Telemetry provider Trillium clocked 431,505 slots during epoch 1021, immediately following the change. The weighted average slot time landed at 365.4 ms, down from 420.7 ms in epoch 1015. Over the same window, the number of skipped slots dropped sharply.
| Metric | Epoch 1015 (before) | Epoch 1021 (after) |
|---|---|---|
| Weighted average slot time | 420.7 ms | 365.4 ms |
| Slots timed | — | 431,505 |
| Missed slots | 1,890 | 331 |
| Missed slot rate | 0.438% | 0.077% |
This drop in missed slots is an encouraging sign for network stability. However, comparing only two non-consecutive epochs does not allow attributing the entire improvement to the timing change alone: other operational factors may also contribute.
Technical impact: budgets, throughput, and leader window
Shortening the slot time reduces the real-time delay needed to reach confirmation thresholds. The four-slot window assigned to each leader shrinks from a nominal 1.6 seconds to 1.4 seconds, compressing the period during which a single block producer controls the sequence.
The proposal keeps four slots per leader and 432,000 slots per epoch. However, the per-slot budgets — compute, account writes, voting, data, and shredding — are reduced proportionally. Blocks arrive more frequently, but with lower ceilings. The per-block compute cap drops to 87.5 million units at 350 ms, down from 100 million units adopted in July. The overall theoretical maximum remains near 250 million units per second.
« In the short term, combining shorter slots with an increased block limit would quadruple the network’s throughput. »
Mert Mumtaz, CEO of Helius
Mert Mumtaz, CEO of Helius, noted on a podcast that pairing shorter slots with a higher block cap could quadruple throughput in the short term, provided the network remains stable through each transition.
Roadmap to 200 ms
The official roadmap plans four successive 50 ms reductions to converge on a 200 ms target: 350 ms (active), 300 ms (pending), 250 ms, then 200 ms. The 300 ms milestone was pending as of August 26. According to Solana Compass, Brennan Watt, CEO of Anza and author of the SIMD, indicated it should take effect at epoch 1024, around August 28.
Each step has its own feature gate, allowing operators to pause the sequence if the missed-slot rate exceeds acceptable thresholds. Feature-gate governance thus contains the risk of a cascade failure during instability.
Testnet and devnet running ahead
Lower benchmarks have already been crossed on Solana’s test environments. The testnet passed 350 ms on August 5 and now targets 200 ms. The devnet sits at 300 ms with the 250 ms feature gate activated but not yet effective. Brennan Watt confirmed that testnet slot times hit 182 ms, with a 193 ms average recorded over the preceding hour. Testnet reductions occur roughly every two epochs.
Mainnet deployment was shipped by Anza, the firm maintaining the Agave validator client, as part of the Agave v4.2 release. One identified challenge concerns off-chain software that still treats the 400 ms slot as a hardcoded constant rather than reading it dynamically from the cluster. Explorers and indexers will need updates to reflect the new tempo.
At the time of the announcement, the SOL token was up 4.37% over 24 hours and held seventh place in the crypto market capitalization ranking, at roughly $45 billion.
Toward genuinely sub-second finality
This first slot-time reduction is a concrete step toward the publicly stated sub-second goal. Early telemetry suggests a stable transition, with no spike in missed slots. Progress, however, will hinge on missed-slot rates at each milestone and on operators’ ability to pause any step when needed.
If the trajectory holds, Solana could within a few months offer a confirmation latency comparable to — or lower than — centralized systems, while preserving its proof-of-stake decentralization properties. The transition from 350 to 200 ms will be the next live test for the ecosystem.
Sources
This article is published for informational and educational purposes only. It does not constitute investment advice in any form. Do your own research (DYOR) before making any decision.
