April 2026 releases // Telegram Go release and more advanced relays

The Go rewrite of Telegram has finally been released. Related to that, there have been multiple improvements to relay mode in bridgev2.

Bridges

Telegram Go rewrite

The Go rewrite of Telegram has been released. Thanks to all the early testers and people who forgot that the :latest docker tag always follows the main branch.

In terms of new features, the highlights are support for topic groups as well as sticker/emoji pack import. Since it's a rewrite, there are a lot of other smaller changes as well, but I'm not going to bother finding all of them. If something is missing, you can bring it up in the Matrix room.

The relay mode in bridgev2 has been extended to allow emulating the old-style relaybots more closely. A bridge command was added to bridge existing remote chats to existing Matrix rooms, which also allows bridging through a default relay user without being logged into the bridge. When doing that, set-relay is called implicitly. Both bridge and set-relay will now default to using default relays if any are configured, even for admins who are logged into the bridge.

Note that there is no migration from the old relaybot to the new relay mode, as the migration doesn't know who you'd assign the login to. I'm not yet certain whether a migration will ever be added to the bridge itself, but it's technically possible to migrate the relevant data manually.

Matrix relays

I've made a proof of concept that allows running any bridgev2 bridge with just a single bot instead of an appservice, using MSC4144 per-message profiles for representing users on the remote network: https://github.com/mautrix/go/pull/484. It still has bugs and it's not intended to replace appservice mode, but once it's cleaned up and merged, it might be useful for running light-weight bridges without a homeserver, e.g. for relaying rooms to many networks.

Other changes

Most other bridge changes were bug fixes and small features. The Messenger bridge has improved support for marketplace chats, including moving them to a separate subspace. See the release notes linked above for more details.

Various crypto bugs were fixed in mautrix-go, bringing goolm closer to being production-ready. Recent testers haven't reported any unexpected undecryptable messages caused by goolm, so switching to it by default later this year is plausible.

The Discord bridgev2 rewrite is progressing, but the current estimate is at least 2 months, likely more.

gomuks

gomuks got a whole bunch of small fixes, as well as a few bigger new features.

There's now an arbitrary profile editor in the devtools/state explorer, which allows you to easily modify any global profile field. Additionally, there's a dropdown to set your pronouns in your own profile view in the right sidebar. However, the pronoun dropdown will only work for simple cases; if you want to use non-English pronouns, use the new arbitrary profile field editor in devtools.

There's a new message context menu item to view the full list of reactions on a given message and delete any of them (including reactions from other users if you have redaction permissions).

The backend and widget API were updated to support sticky events, which should future-proof the Element Call widget for when they switch to sticky events. It hasn't actually been tested yet though, so it's possible it'll end up exploding.

Finally, setting up cross-signing and key backup is now supported. When logging in, gomuks web will detect whether the account already has key backup set up and offer appropriate buttons based on that. New accounts can set up keys, while old ones can reset keys if the recovery key was lost. This means that on servers using SSO or next-gen auth, you should be able to register and fully set up an account with only gomuks web. The legacy registration API is not supported though, so for such servers the account still has to be created separately.

Meowlnir

Meowlnir only had some minor improvements to the policy server, plus support for obfuscating profiles when banning users to prevent malicious user IDs from sticking around in the timeline. If you do use policy servers, the new cache for signatures is somewhat important, as the database can get overloaded otherwise.