SIMs, Carriers, and What the Data Actually Costs

Introduction

A gateway with no service plan is an expensive paperweight, and this is the part of the decision that is least discussed in reviews and most likely to produce an unpleasant surprise. The hardware question has a clean technical answer. The plan question is a moving target governed by carrier policy rather than physics, and policy changes without notice.

4.1 The carrier choice is a coverage decision, and it is not close

For the Upper Peninsula specifically, the carrier ranking is not the same as the national one. Crowd-sourced measurements of Keweenaw County consistently place AT&T first, Verizon second, and T-Mobile third by covered land area — the reverse of what a metro-area speed test would suggest.

That has a direct consequence for how the two SIM slots should be filled. The dual-SIM capability is worth more in the Keweenaw than the antenna is, because it addresses coverage rather than reception, and coverage is the binding constraint. Two SIMs on the same carrier is a wasted slot. The sensible configuration is:

• Primary: AT&T, for the best chance of any service at all in the Upper Peninsula.
• Secondary: T-Mobile, chosen deliberately for n71 / B71 at 600 MHz — the lowest-frequency allocation in wide use, and the one most likely to reach into forested terrain where AT&T’s 700 MHz does not. Where T-Mobile has built out at all in the UP, it tends to be low-band, which is exactly the propagation characteristic this build needs.

The two are complementary rather than redundant, and that complementarity is the actual argument for a dual-SIM gateway over a single-SIM one.

Verizon is the awkward third. It measures competitively in the region but has historically been the most restrictive about activating service on devices it did not sell, which is the subject of the next section.

4.2 The device-certification problem

This is the part that bites people.

Carriers maintain allowlists of device IMEIs. A SIM moved from a phone into a third-party router may work, may work until an audit, or may refuse to activate at all — and the behaviour differs by carrier and by plan. Consumer unlimited plans in particular are written for phones, and using one in a router is generally a terms-of-service violation even when it technically functions.

The GL-X3000’s specific advantage here is that it is certified by both T-Mobile and AT&T as an IoT device. That certification is not marketing: it means the IMEI is known to those carriers and the device can be provisioned on their IoT and data-device plans through the front door rather than by accident. Given the AT&T-primary, T-Mobile-secondary recommendation above, the two carriers this router is certified on are exactly the two it should be used with — which is a genuine point in the product’s favour and one worth weighing against a competitor that is not certified.

4.3 A practice to avoid

The community workaround for restrictive plans is to change the router’s IMEI to that of a tablet so the carrier’s systems see an approved device and apply a cheap tablet data line. It is widely described in forums and it does work.

It should not be done here. Setting aside that altering a device identifier to obtain service by misrepresentation is a terms-of-service violation that can and does get lines terminated without refund — usually at the least convenient moment, which for this use case means in the Keweenaw — the legal status of altering an IMEI in the United States is not something to be casually assumed benign. There is a legitimate path available on both of the carriers this device is certified for, and the money saved does not justify a line that can vanish mid-trip.

4.4 What a plan realistically costs

Precise figures change constantly and should be re-checked at purchase, but the shape of the market is stable enough to plan against:

Table 1 — Precise figures change constantly and should be re-checked at purchase, but the shape of the market is stable enough to plan against

Path	Rough monthly	Character
Carrier IoT / data-device plan (AT&T, T-Mobile), metered	roughly $20–40 for a modest bucket	The clean, sanctioned option. Predictable, survives audits
Carrier “connected device” unlimited-ish plans	roughly $50–70	Usually deprioritized behind phone traffic and often speed-capped after a threshold
Adding a data line to an existing phone account	often $10–20	Cheapest if the account supports a data-only device; check whether the router is an eligible device before assuming
Prepaid, activated seasonally	varies	Worth serious consideration here. This is a trip-driven use case, not a daily one

That last row deserves emphasis. A rig used for a handful of trips a year does not need twelve months of service on two SIMs. A prepaid or suspendable line on each carrier, activated for the weeks it is needed, can cut the annual cost by more than half — and it changes the economics of the whole decision, because a recurring $60 a month for a capability used six weeks a year is $720 for six weeks of internet.

4.5 Data consumption, honestly

Overlanding data use is generally modest and bursty, which argues against buying a large bucket:

• Weather, forecasts, and radar before committing to a route — kilobytes to a few megabytes.
• Messaging and email — negligible.
• Map downloads — the one large item, and the one that should never happen in the field. Gaia GPS caches offline, and downloading the region before leaving pavement is the correct workflow regardless of what the router can do. Any trip where maps are being pulled over cellular is a trip where the preparation failed.
• Video streaming in the tent — the item that dominates any bill it appears on, and the honest reason most people buy this equipment.

The single most valuable configuration change is to enable the data-limit and per-client controls in the GL.iNet interface so that a background OS update on a laptop cannot consume a metered bucket in an afternoon. This is a five-minute setup step that prevents the most common way these plans go wrong.

4.6 The failover behaviour, stated plainly

The GL-X3000’s dual-SIM is failover, not concurrent. One modem, one active SIM. When the active carrier drops, the router switches to the other slot, and that switch takes on the order of tens of seconds and drops any session in progress. A video call will end. A file transfer will restart.

This is the correct behaviour for the use case — an interruption is vastly better than an outage — but it should not be mistaken for the seamless bonded failover that professional gear provides, and it means the primary SIM choice still matters even with two installed.