Flint 4 BE-14000 Discussion Thread

Hi,

We understand that the current Wi-Fi + WireGuard performance can be disappointing.
We have shared this behavior with the relevant team, and they are continuing to investigate and explore possible optimizations for the Wi-Fi-to-VPN forwarding path.

If there are any new findings or further progress, we will update the thread accordingly.
Thank you for your patience and support.

If i were to plug in my 10G internet into Flint 4, for maximum performance, should i plug in SFP+ 10G BaseT or the 10Gbps local port?

DNS hijack rules are only generated for lan/guest/iot — custom VLAN bridges leak in cleartext (4.9.1)

GL-BE14000, firmware 4.9.1. Reporting a DNS leak that the web UI gives no indication of.

Setup: WireGuard client with a domain-based VPN policy, AdGuard Home enabled (Handle client requests OFF), and a custom VLAN interface vlan59 added via LuCI with its own firewall zone.

Issue 1 — Override DNS Settings for All Clients does not cover custom bridges.

With that option enabled, the generated PREROUTING rules are:

-A PREROUTING -i br-lan -p udp --dport 53 -j dns_dispatcher
-A PREROUTING -i br-guest -p udp --dport 53 -j dns_dispatcher
-A PREROUTING -i br-iot -p udp --dport 53 -j dns_dispatcher

br-vlan59 is absent, even though it has a proper firewall zone in /etc/config/firewall. Three devices on that VLAN with hardcoded 8.8.8.8 were sending cleartext DNS straight out the WAN — 654 concurrent conntrack entries, SNAT'd to my real WAN IP. AdGuard Home never saw them, and neither did the tunnel. Nothing in the UI indicates the option's coverage is limited to the three built-in zones.

Issue 2 — AdGuard Home upstream DoT intermittently egresses via WAN.

With Services from GL.iNet Use VPN irrelevant here (AdGuard runs as gid 20000 and is matched by TUNNEL*_LOCAL_POLICY), the marking rule works — 310 packets matched. But long-lived DoT sockets created before the policy loads keep their original route. Measured 4 of 13 upstream connections to 1.0.0.1 / 8.8.4.4 egressing via WAN with the real IP, hours after boot. The per-tunnel Kill Switch does not catch this: its DROP rule sits in the same chain behind a MARK rule that always matches first, so its counter stays at 0.

Note for anyone writing their own guard rule: matching on ! -o wgclient1 in filter OUTPUT does not work. At that point the packet still carries the pre-reroute output interface, so the rule matches everything and takes DNS down completely (I did this — 285 rejects, total resolution failure). Match the fwmark instead:

iptables -A output_rule -p tcp --dport 853 -m mark ! --mark 0x1000/0xf000 -j REJECT --reject-with tcp-reset

Hi,

Thank you for the detailed report.

Regarding the custom VLAN:

We reproduced the behavior with a custom network created through LuCI.

The network was not automatically added to the DNS rules managed by the GL.iNet web Admin Panel, so DNS requests sent directly to an external DNS server could bypass the DNS override.

However, Networks created through the GL.iNet web Admin Panel are automatically registered and included in the corresponding DNS rules.
Therefore, we recommend creating and managing the custom network through the GL.iNet web Admin Panel.


If you prefer to create the network through LuCI, the corresponding TCP and UDP DNS redirect/routing rules need to be added manually.

Regarding the DoT traffic:

We tested this locally but were unable to reproduce DoT payload continuing through the WAN after the VPN was fully connected.

In our test, the WAN interface only showed FIN retransmissions from the previous connections, while the subsequent DoT connections and payload were observed on the WireGuard interface.

If the issue can be reproduced again, could you please provide the system log covering the router startup and VPN connection, together with short WAN and WireGuard captures from the same time period and send the files to us by private message?

Thank you for your understanding and support.

Hi GL.iNet team, I am seeing a very similar Wi‑Fi + WireGuard performance issue on my Flint 4 (GL-BE14000, firmware 4.9.1). I have already contacted support and tried the suggested settings, but the problem remains. Could this case be escalated to the engineering or senior technical support team?

My client is a MacBook Pro M5 Max, and I use ProtonVPN WireGuard. My approximate download results are:

Connection Download speed
Ethernet, router VPN off 4,000–4,300 Mbps
Ethernet, router VPN on Around 1,000 Mbps
Wi‑Fi, router VPN off Around 1,600 Mbps
Wi‑Fi, router VPN on Around 400–540 Mbps
Same Mac on Wi‑Fi, using the ProtonVPN app on the Mac instead of the router VPN Around 934 Mbps

I have tried changing the Wi‑Fi settings, including using 5 GHz and 6 GHz for MLO, and changing the hardware acceleration and related settings. I also tried MTU 1380. None of these resolved the Wi‑Fi slowdown through the router’s VPN client.

The Mac can reach around 934 Mbps when it handles the VPN itself over the same Wi‑Fi connection, while the Flint 4’s WireGuard client gives me only around 400–540 Mbps over Wi‑Fi. This seems close to the issue reported in this thread. I also experienced a similar pattern with a UDR7, so I would appreciate help identifying exactly where the bottleneck is rather than assuming a cause.

Could the engineering team advise:

  1. Whether this Wi‑Fi-to-WireGuard forwarding slowdown on firmware 4.9.1 is a known issue?

  2. Whether there is a firmware fix, test build, or setting that could improve it?

I bought the Flint 4 specifically for fast VPN performance, so I would really appreciate a technical investigation. Thank you.

The Wi‑Fi-to-WireGuard performance difference has already been shared with the relevant team for investigation, as confirmed earlier in this thread. However, there is currently no confirmed firmware bug, fix, test build, or additional setting announced for firmware 4.9.1.

Your same-device comparison—approximately 934 Mbps with the ProtonVPN app versus 400–540 Mbps through the router’s WireGuard client—is useful evidence. Please add these results and a link to this thread to your existing support conversation so the complete case history can be reviewed together. Do not post your support case number, logs, or other private details publicly.

This reply was generated by AI. Please verify its accuracy.

GL.iNet Technical Support reviews and responds to relevant threads on business days. Please allow time for a staff response.

Hi its now acknowledged as a known issue. A temporary workaround is to set the rps_CPU value to E which balances the wireguard encryption across the other CPU cores instead of CPU1 which the router is defaulting to. See the latest posts in the ‘Meet Flint 4’ topic. Particularly input from user @Void who helped diagnose and come up with the workaround. We all eagerly await a proper firmware fix, as well as updated kernel. Lord knows why this router shipped with openwrt 21 ! Its 2026!!!

It launched with openwrt 21 because that is what MediaTek support in their sdk.

It is completely normal that in 2026 a device using a MediaTek chipset would be use openwrt 21. That is what MediaTek forked and maintain for their current official sdk that they support.

MediaTek's official SDK still ships a 21 branch but it's no longer the only one. Per their repos they now maintain 3 in use branches inc 24

Thanks for posting the temporary solution here and the testing.
So it is a real problem with the firmware and now we hope they bring asap a proper firmware solution as part of an update.
Only thing I not understand is why, when connected directly over an ethernet with Wireguard connection, this problem not shows up?
I would assume that it would also be a propblem when connected over ethernet too.
So probably the Gl.iNet guys will figure it out soon because it is like this since 24 days ago when I received the device.

I’d also like to hear from other Flint 4 owners: are you seeing a similar difference between Ethernet and Wi‑Fi when using the router’s WireGuard client?

Has anyone actually resolved the Wi‑Fi-specific slowdown by switching VPN providers or WireGuard servers, without changing rps_cpus or other router settings? If so, which provider did you use, and what were your approximate Ethernet-VPN and Wi‑Fi-VPN speeds before and after switching?

I’ve seen the reports that changing rps_cpus improved Wi‑Fi throughput, which suggests the router’s packet processing may be part of the issue. I’m not comfortable making SSH changes myself, so I’d particularly value results from people using the standard firmware and settings. Please don’t post VPN configuration files, logs, or IP addresses just to answer this.

Has anyone in this thread tried to install beta V4.11.0 ?

It comes with OpenWRT 25.12 as far as I can see. Can‘t try to install, as I am on vacation right now :smiley:

Several users have installed Flint 4 firmware 4.11.0 beta and are reporting results in the SDK v4.11 beta thread. It is beta firmware, and Flint 4 users have already reported some issues, so it should not be treated as a stable upgrade.

The version number alone does not confirm that it is based on OpenWrt 25.12. Someone running it can verify the actual base and kernel over SSH with:

cat /etc/openwrt_release
uname -r

This reply was generated by AI. Please verify its accuracy.

GL.iNet Technical Support reviews and responds to relevant threads on business days. Please allow time for a staff response.

I received an firmware update to 4.9.2 today but it contained no solution for the bad Wifi throughput with Wireguard.

I have a new Flint 4 (GL-BE14000) being used as my main router, with Flint 3 units as mesh nodes.

The router has experienced a complete loss of internet/network connectivity. Wi-Fi was still visible/connected, but devices had no internet access.

I checked the system log immediately afterwards and found these errors at the time of the failure:

mtk_soc_eth 15100000.ethernet eth0: NETDEV WATCHDOG: CPU: 3: transmit queue 3 timed out 5050 ms

mtk_soc_eth 15100000.ethernet eth1: NETDEV WATCHDOG: CPU: 3: transmit queue 1 timed out 5060 ms

The timeout then continued/increased, and the log also reported:

detect QDMA Tx hang !

Both eth0 and eth1 appear to have stopped transmitting at essentially the same time.

There are also many wappd messages immediately before the failure, including:

Get Beacon report is invalid.

The router itself did not appear to completely reboot or kernel panic — it looked more like the Ethernet/network subsystem became stuck while the router continued running.

Has anyone else seen this on the Flint 4 / BE14000?

Could this be a MediaTek Ethernet/QDMA driver or firmware issue?

Those messages indicate that the MediaTek Ethernet transmit queues stopped making progress in the QDMA path. That fits the observed loss of both WAN and LAN forwarding while Wi‑Fi remained associated. However, the log excerpt alone cannot establish whether the underlying cause is the Ethernet driver, hardware acceleration, firmware, or another component; the preceding wappd beacon-report messages may be unrelated.

Please provide the exact Flint 4 firmware version and build date, and state whether Network Acceleration was set to Hardware, Software, or None when this occurred. If it happens again, export the system log before rebooting and send it privately to a GL.iNet technical support member, together with the approximate failure time and which physical ports were in use. Do not post the complete log publicly. As a targeted comparison, temporarily set Network Acceleration to None and observe whether the failure recurs; that result would help determine whether the accelerated forwarding path is involved.

This reply was generated by AI. Please verify its accuracy.

GL.iNet Technical Support reviews and responds to relevant threads on business days. Please allow time for a staff response.

I have loaded up the beta Version 4.11.0 Firmware Type beta1 and am quite happy with it (now).

Originally I made the mistake of using the apk command to updated and to upgrade - that was a big mistake and stuffed everything up! My recommendation is not to do the offered updates/upgrades.

I have a Flint 4 main with a Flint 4 MESH node on ethernet. I am using a SFP 2.5 gig ethernet module to connect to the internet on the main router. I am using FWIW, the 10 gig port to connect the Flint 4 MESH Node.

I am using AdGuard + Unbound and I have a docker container with SearXNG on the Node. I have configured WireGuard and regularly connect a TAB 10 to my mobile phone (tethering) data and use WireGuard to connect to home.

My needs are simple and I do not go looking for issues with my system seeming to be working fine.

I like the touch screen but do concede that once I put the router in the final position up on a shelf, it will not be used.

I like the features of the Beta, MESH, The network Quality GUI, security scan (and of course the usual AdGuard and the ability to use docker).

I am not using a FLINT 3 as a node but my internet is "rock solid":-

(I hate that GL.iNet have resorted to using AI to post in the forum - I think it is just normal AI slop!)

Update after upgrading my Flint 4 to firmware 4.9.2

I wanted to share some good news and thank everyone who helped investigate this issue, especially the GL.iNet technical support team.

After upgrading my Flint 4 (GL-BE14000) to firmware 4.9.2, I tested the connection again and saw a clear improvement in my router VPN result. In my latest Speedtest, ProtonVPN WireGuard on the router reached 1,159 Mbps download and 433 Mbps upload. For comparison, running ProtonVPN directly on my MacBook Pro reached 1,168 Mbps download and 245 Mbps upload in a separate test.

Previously, I was seeing only around 400–540 Mbps download when using the router’s VPN over Wi‑Fi. The latest result is much closer to what I had hoped for when I bought the Flint 4. I will continue to use it and see whether the improvement remains consistent.

Thank you to the community members who shared their findings, and to the GL.iNet technical support team for following up. I appreciate the work that has gone into improving the firmware.

Has anyone with a Flint 4 reached around 1.5 Gbps over WireGuard? If so, which VPN provider are you using? I’m getting around 1.2 Gbps with Proton VPN and wondering whether that’s maximum?

Finally, May i know that anyone upgrade to 4.11.0 beta, how about the Network Quality and Security Scan?

The advertised up to 1.5 Gbps WireGuard speed is a laboratory maximum measured with a wired client, not a guaranteed result over Wi‑Fi or with every VPN provider. Your 1,159 Mbps result is already close to that reference; actual throughput depends on the VPN server, route, encryption load, client connection, and test server. For meaningful comparisons, use the same client and Speedtest server at similar times.

In firmware 4.11.0 beta, Network Quality reports metrics such as latency, jitter, packet loss, throughput, speed-test results, and bufferbloat grade. Security Scan checks areas including Wi‑Fi security, WAN exposure, application vulnerabilities, and firmware updates. Community experiences are being collected in the SDK v4.11 beta thread; because this is beta firmware and Flint 4 issues have been reported, it should not yet be treated as a stable upgrade.

This reply was generated by AI. Please verify its accuracy.

GL.iNet Technical Support reviews and responds to relevant threads on business days. Please allow time for a staff response.

Um..... I do not understand the question... I posted my image of the Network quality above
Security scan shows:-