EQUITYAAPL$349.51▲5.31%
EQUITYNVDA$200.13▼3.17%
CRYPTOBTC$80,589.42▲4.88%
EQUITYTSLA$385.19▼3.34%
ATLASMarket intelligence
Atlas · MARKET NOTE
Published 2026-11-05

Sui's 4 November 2026 mainnet release tightens transaction limits and node responses

Sui published mainnet-v1.83.1 on 4 November 2026. The release moves the Sui Protocol to version 141 and combines a protocol-level transaction change with more explicit behaviour from validators, full nodes and data APIs. The practical theme is narrower tolerance for ambiguous or oversized requests: some failures are decided earlier, and some incomplete or out-of-range reads now return a named error rather than a misleadingly tidy response. Read the mainnet release.

What changed

At protocol level, version 141 refines the transaction-effects size estimate and limits a transaction to 1,024 accumulator events. The release also includes bulk vector operations such as replace_range, keep_range, copy_range, remove_range and truncate, alongside support for references returned from Move functions in programmable transaction blocks.

The same release notes say that protocol version 140 enabled two mainnet behaviours. Object-fund withdrawals that exceed the available balance now fail during execution with the EObjectFundsInsufficient Move abort, rather than through a post-execution InsufficientFundsForWithdraw error. Transaction effects also no longer report transaction dependencies, while retaining an empty dependency vector and leaving object versions and per-object previous-transaction metadata unchanged.

Those details matter because they affect what developers can observe and how they handle failure. An application integrating with Sui may need to distinguish an execution-time funds failure from an error found after re-execution. A client consuming transaction effects may also need to account for dependency data that is structurally present but empty.

The release tightens node and data-service behaviour as well. Validators now cap the user-transaction bytes they propose to consensus, with a default of two full blocks per second. Operators can change that setting or disable the rate limit. State sync can download checkpoint contents in pages when a full download does not complete, while tidehunter nodes can retain up to 2 GiB of hot objects-index blobs in memory after they age out of the memory-mapped window.

For gRPC clients, SimulateTransaction now returns structured ErrorInfo when gas selection cannot find enough SUI. That response can include the owner, required budget, coin balance and address balance, so clients do not need to parse a free-form error message. The release also changes several below-retention or beyond-index requests from INTERNAL, an empty page or a clamped result to OUT_OF_RANGE. See the gRPC and node changes in the release notes.

GraphQL receives a similar precision pass. Nullable, fallible multi-gets can return successful entries alongside indexed errors instead of failing the entire batch. Subscription resumption no longer skips a checkpoint at the handoff from backfill to live data, and object and package resolution in several transaction results is corrected when an object was not written by the requested transaction or checkpoint.

“Sui is making the edges of its data plane more legible: when a request cannot be satisfied, the answer is increasingly a named boundary rather than a plausible-looking shrug.”

Why it matters

This is a release about contract clarity more than a disclosed commercial milestone. Better error metadata can reduce the amount of client-side guesswork around gas simulation. Explicit OUT_OF_RANGE responses can help indexers distinguish “there is no matching item” from “the requested position is not currently available”. The transaction-event limit and proposal-byte cap place concrete boundaries around workload shape and consensus intake.

That does not establish faster applications, greater transaction demand or improved network economics. The source pack provides implementation details, not measurements of usage, reliability, developer adoption or asset performance. It also does not establish that every client has already adapted to the changed responses.

There is a useful counterpoint in the chronology. Sui published testnet-v1.83.1 on 3 November 2026, one day before the mainnet release, and its notes describe protocol version 140 and many of the same node, gRPC and GraphQL changes. The mainnet release then identifies protocol version 141. Read the testnet release. The supplied extract does not provide rollout metrics or a comparison of live behaviour across the two networks.

Market Wit

Sui has spent this release teaching its infrastructure to say “that request is out of bounds” with more confidence. In a system built to process programmable transactions, even the error messages now have a little governance: fewer polite empty pages, more constitutional limits.

The open question is operational rather than theatrical. Do applications treat these new structured responses as durable interfaces, or do they still depend on parsing the wording around them? The next check is the relevant client and node documentation, plus live integration behaviour after the 4 November 2026 mainnet release.

Explore the underlying Atlas research

FreedomCore Atlas Research →