Cheap Router, OpenWrt, and a Mesh Kit Walk Into a Lab

Most cheap routers are not bad hardware. They are bad software with a plastic shell. That is the premise behind this month-long test: take a $40 dual-band router, flash it with OpenWrt, and see whether it can replace a $199 three-node Wi-Fi 6 mesh kit in a real US home office. We ran the comparison from mid-October through mid-November 2026 in a 900-square-foot office with six active clients, one wired NAS, two video-call-heavy desks, and a garage office that has always been a Wi-Fi graveyard.

The hardware and the rules

The bargain box was a $40 dual-band AC router from a big-box sale, the kind with four Ethernet ports and a power brick that feels too light. We flashed the current stable OpenWrt release and added no paid packages. The mesh kit was a mainstream $199 three-node Wi-Fi 6 system, set up through its phone app with default settings plus a guest network. Both shared the same 500 Mbps cable connection. We rotated them every other week to avoid blaming one bad ISP afternoon on the wrong box.

We tested wired throughput, close-room Wi-Fi, far-room Wi-Fi, bufferbloat under a saturated upload, WireGuard VPN speed, guest isolation, update behavior, and power draw at the wall. This was not a spec-sheet shootout. It was a month of 'can I take this call without sounding like a robot?'

Setup friction: the CLI tax is real

The mesh kit was online in 12 minutes. The app found the nodes, asked for a Wi-Fi name, and offered a firmware update. My mother could do it. Probably.

OpenWrt took 2 hours and 20 minutes, including the part where I checked the router's exact hardware revision twice because the first firmware image was for a sibling model. Flashing itself was quick. The slow part was deciding what I actually wanted: VLANs, a guest SSID with client isolation, SQM for bufferbloat, and a WireGuard tunnel back to a friend's lab. OpenWrt can do all of that. It just asks you to mean it.

That is the first real cost. The $40 router is not a $40 project if you bill your own time.

Throughput: the mesh kit wins the open floor, OpenWrt wins the weird corners

Wired throughput was close enough to call a tie: about 870 Mbps through the OpenWrt box and 910 Mbps through the mesh kit's main node. Wi-Fi in the same room told a different story. The mesh kit hit roughly 620 Mbps on 5 GHz; the OpenWrt router managed about 410 Mbps. Two walls away, the mesh kit held around 240 Mbps. The cheap router dropped to 85 Mbps and occasionally sulked.

But the OpenWrt box had one advantage the mesh app could not match: I could put it exactly where I wanted and aim its radios. In the garage office, a $12 flat Ethernet cable and a second access point would beat either standalone kit. If you are comparing one router against three mesh nodes, the mesh kit should win coverage. It did.

Bufferbloat: the cheap router's best party trick

This was the surprise. With SQM and cake shaping enabled, the OpenWrt router kept our connection at an A grade for bufferbloat during a full upload saturation test. Video calls stayed usable while a backup job chewed through the line. The mesh kit, with its default settings, slid to B and sometimes C. Its app offers no real bufferbloat controls.

For a small office where two people are on Zoom and one is uploading a client file, that matters more than a peak speed number. OpenWrt is not magic; you have to configure SQM for your actual upload speed. But once it is set, the line feels calmer.

VPN performance: CPU matters

WireGuard on the OpenWrt router topped out around 180 Mbps in our test, which is plenty for remote file access and fine for a small office. The mesh kit's built-in VPN was easier to turn on but slower, around 70 Mbps, and offered fewer routing options. If you need site-to-site or split tunneling with real rules, OpenWrt is the more capable tool. If you need 'turn on VPN, hand phone to sales guy,' the mesh app wins.

Guest network isolation: one toggle versus a firewall lesson

The mesh kit's guest network took one toggle and isolated guests from the LAN. OpenWrt took a firewall zone, a separate SSID, and a DHCP range. It worked, and it gave me more control—I could allow access to one printer and nothing else. But I also broke it once by forgetting to allow DNS. That was a 15-minute detour caused by confidence, not software.

Updates, power, and the boring stuff

OpenWrt gets security updates regularly, but you have to pay attention. Attended sysupgrade makes it less painful than it used to be, yet it is still a maintenance task. The mesh kit auto-updated twice during the month without asking. That is convenient. It is also a small trust exercise.

Power draw was not close. The OpenWrt router pulled about 6 watts at the wall. The three mesh nodes together pulled about 18 watts. At a US average of roughly 17 cents per kWh, that is about $18 per year saved. Nice, but not a reason to choose a router.

The break-even point

Here is the math that matters. The hardware gap is about $159. OpenWrt cost me 2.3 extra hours to set up. If your time is worth $50 per hour, that is $115, so you are still ahead in year one. If your time is worth $75 per hour, the setup tax is $172.50 and the mesh kit wins on day one. Add an hour a year for updates and the cheap router's advantage shrinks further.

So the break-even is not really about years. It is about whether you value control or convenience. If you want SQM, granular guest rules, WireGuard, and the ability to keep a router out of the e-waste pile, the $40 OpenWrt box is a legitimate small-office tool. If you want three nodes, an app, and automatic updates, pay the $199 and go do actual work.

We kept the OpenWrt router on the main line after the test. Then we plugged the mesh kit back in for the family room, because sometimes the best answer is both.