// what I verified
Claims, and where each one stands.
Every claim on this page carries its evidence tier. Nothing is smoothed over, and nothing a project told me is presented as something I checked.
Assets stay in the user's own Canton account and the operator takes no custody. Users hold their own keys.
cantex.io, docs.cantex.io termsMatching happens off the ledger, then each trade settles atomically on Canton in a single on-ledger transaction. There is no partial-fill exposure and no reconciliation step.
cantex.io, docs.cantex.ioBuilt and operated by CaviarNine Limited, registered in the British Virgin Islands, founded by Oliver Scott-Simons and Chris Colman, both from derivatives and electronic markets backgrounds. The team previously shipped on-chain products on Radix.
docs.cantex.io terms and team pagesThe AMM is live. A public Python SDK is published under an open licence on GitHub, and a direct Ethereum bridge is running.
github.com/caviarnine/cantex_sdk; CCTools campaign pageThird-party assets list on the venue, including HECTO and EDELx. Cantex is not a single-asset venue.
@Cantex_io announcementsCantex states it has passed $100M in total trade volume. Self-reported on its own channel, with no methodology, date range or third-party verification published alongside it.
@Cantex_io on XA high-throughput limit order book with a Rust matching engine is described as Phase 2. No launch date has been published. On 2 August 2026 Cantex told me directly that the order book is unlikely in the near term. That is the project withdrawing its own announced item, which is the one correction type where the project is the authoritative source. The item stays on this page rather than disappearing from it.
cantex.io; docs.cantex.io roadmapEdel Finance states its Runway partnership with Cantex saw daily trading volume approach $100,000 at its peak. This is Edel's figure from Edel's release, describing that partnership rather than the venue.
Edel Finance press release, 9 July 2026Third-party directories describe Cantex as Canton's primary on-chain trading venue. That is an editorial characterisation, not a market-share measurement, and no comparable public volume data exists across venues to test it.
canton.wiki editorial; cantonnews.orgMarketing copy describes an engine capable of hundreds of thousands of orders per second. That is a stated design capability for a component not yet live, not a measured throughput figure.
cantex.io// what would move it
Catalysts.
Each one names what I am watching and what would prove me wrong.
The AMM serves swaps. An order book is what professional desks actually need, and it is the difference between a venue retail uses and a venue institutions route through. Announced as Phase 2, undated, and the single most consequential item on the roadmap.
The $100M figure is self-reported with no methodology attached. A venue that publishes verifiable volume at a stable location changes what anyone can responsibly say about it, including me. Until then the number stays announced-tier no matter how often it is repeated.
// what I could not verify
Gaps.
These are findings, not omissions. I looked and state what I found.
Third-party security audit. No published audit was found as of 22 July 2026. The absence of a public audit is not evidence of a problem, and it is also not evidence of safety.
Volume methodology. The $100M total is stated without a date range, an inclusion rule, or a way to check it. No liquidity, TVL or user figures are published at all.
Order book timing. No date, quarter or milestone was ever attached to Phase 2. As of 2 August 2026 the project states it is unlikely in the near term, and no replacement roadmap item has been published.
Canton Quadrant. Cantex was not scored in issue 01, where it appeared on the radar list. It was scored in issue 02 at 3 of 7 confirmed substance, on S1 live deployment, S3 official ecosystem status and S5 legal entity with named leadership, at token intent 0. S2, S4, S6 and S7 scored zero. This page predates the checklist block used on pages built for issue 02, so the item-level detail lives in the issue rather than here until the page is rebuilt.
// open to the project
Sections Cantex can fill.
These stay empty until the project supplies them. Anything supplied is labeled as supplied, dated, and never rendered as something I verified. Declining changes nothing else on this page.
Architecture detail
How matching, signing and settlement actually fit together, in the team's own words.
Audit and security posture
Any audit, review or internal security practice not published today.
Volume methodology
What the $100M total counts, over what period, so the figure can move off announced tier.
Roadmap with dates
What ships next and when, so the catalysts above can carry real dates.
I run campaigns with Cantex. This page was not part of the deal.
Canton Catalyst has run two access code campaigns with Cantex, and that relationship is disclosed on every page it touches. Partnership buys distribution, never coverage. This page meets the same evidence standards as one about a project I have never spoken to, and nothing on it can be bought, thinned or withheld as a consequence of any commercial decision.
Position disclosure. I hold $EDEL and $HECTO, both of which list on this venue. I hold no position in Cantex or CaviarNine.
// corrections
Correction log.
2 August 2026 · announced item withdrawn by the project. Cantex confirmed that the limit order book announced as Phase 2 is unlikely in the near term. No score or tier changes, since the item was already at announced and announced scores zero. The catalyst and the gap are updated to record it, and the claim is retained rather than removed so the roadmap history stays visible. The project also indicated other developments are in progress. Nothing is recorded on that until something is published.
First published 22 July 2026.