Four Free Container Load Calculators, Tested on Ten Loads

Each load was entered by hand into the free tier of every tool we could reach without an account or a payment, on 7 and 8 October 2026, in Chromium 151 at 1,280 px wide. Sizes went in as centimetres (our millimetres divided by 10, with one decimal where needed) and masses as kilograms per piece. “May turn any way” was entered as each tool’s free rotation and “this side up” as its upright setting; where a tool has no such setting, the results say so.

Each tool used its own preset for the container named — 20 ft, 40 ft, 40 ft high cube, 40 ft high cube reefer — because a reader would not type a custom box either. The presets differ a little from ours and from each other: the inside heights of the dry boxes agree within a centimetre, but payloads for a 40 ft high cube are up to 2,100 kg apart, and reefer heights run from 243 to 255.4 cm. So each volume percentage is against the tool’s own container, and weight percentages cannot be compared across tools at all.

Times run from the action that starts a calculation to the result on screen, by wall clock on one machine. containerloadcalculator.com calculates as its page opens, so its times include the page load. Gateway Lines draws its packing as an animation; it was run in a visible browser with its fast-forward button pressed, which changes how fast the plan is drawn but not the plan.

Our planner ran on the same ten lists outside a browser, with two settings changed from its page defaults to match what the other tools do: full support under every piece (the page defaults to 70 %, which suits shrink-wrapped units and usually loads more), and what does not fit going into another container of the same type. Its improvement search was given 1 s per container, where the page gives it 0.6 s. The time in our column is that budget, not a measured page time.

Each cell is the tool’s own reading: pieces loaded of pieces given, the containers used, and for each container the pieces, the share of its volume and, where the tool gives one, the share of its payload. On five of the ten loads — the EUR pallets, the US pallets, the machinery, the tile pallets and the three stops — all four tools put the same number of pieces in the first container, within two tenths of a point of volume. The other five are where they part.

Gateway Lines plans one container. Where a load needs two, it lists the rest as not fitting instead of planning it, which is why its loaded count is lower on five loads; that is a design choice rather than a packing result. On the cartons-and-pallets load its single 40 ft high cube holds 83.0 % of its volume, more than any tool’s first container.

Three loads measure packing alone. On the single carton size in a 20 ft, no plan can hold more than 292 cartons: lengths of 600 and 400 mm add up to at most 5,800 mm within the 5,898 mm floor, and to at most 2,200 mm within the width and the height, so at most 5.8 × 2.2 × 2.2 m of carton fits. Our planner put 270 in the first container, about 92 % of that bound; containerloadcalculator.com 262, about 90 %; cbm3.net 258 and Gateway Lines 255. Gateway Lines cannot lay a carton on its side: its rotation setting swaps length and width only.

On twelve carton sizes in a 40 ft high cube, containerloadcalculator.com filled 95.5 % of its first container and ours 95.4 % of ours, which is 0.8 cm taller. That is level. cbm3.net filled 80.2 % of its first container and 28.4 % of its second: on this load it did not fill the first container before starting the next. Gateway Lines accepts ten lines at most, so it was given ten of the twelve, and filled 84.8 %.

On cartons with EUR pallets that may carry nothing, containerloadcalculator.com put 409 pieces in the first container and ours 408, at 67.5 % and 65.4 % of the volume; cbm3.net put 152 there, at 54.5 %.

Our improvement search stops on a clock, so a faster or slower computer can change a plan slightly. Two runs of the twelve-carton load on the same machine put 1,133 and 1,167 pieces in the first container, both at 95.4 %. The table shows the first run.

containerloadcalculator.com’s plans were also checked geometrically, not only on screen: its packer is a plain JavaScript module, and running it on the same inputs gives every piece’s position. In every container it used, no piece overlaps another, sticks out of the box, stands on less than its full base, or stands on a piece marked not stackable or on a lighter one. The one exception is the tall crate, which is taller than its own door. cbm3.net and Gateway Lines were judged on their 3D views, where we saw no floating or overlapping pieces.

The door. The tall crate is 2.32 m high: under the 2.39 m roof of a 20 ft, over its 2.28 m door. Gateway Lines leaves it out and names the reason, “door opening”, and so does our planner. containerloadcalculator.com loads it without a warning; its packer compares a piece with the inside of the box and never reads the door size it stores. cbm3.net loads it unless the user types the door size under its equipment restrictions; with the size typed in, it leaves the crate out and notes “No definitive physical reason was established.”

The reefer load line. The reefer load is 900 cartons 30 cm high in a 40 ft high cube reefer. Eight tiers reach 2,400 mm. That is under the red line in some boxes — Hapag-Lloyd publishes 2,420 to 2,474 mm, Maersk 2,415 mm — and 6 mm over it in CMA CGM’s, at 2,394 mm. Our planner uses the lowest, so that a plan holds in any of the three, and loads seven tiers: 728 cartons in the first container, against 832 for containerloadcalculator.com and 848 for cbm3.net, which load eight. Neither has a load-line setting, and Gateway Lines offers no reefer. Part of the gap in that row is our cautious line, not packing.

Delivery order. The three-stop load has 120 cartons per stop. containerloadcalculator.com has no stop field and loads lines in the order typed, so with stop 1 listed first, stop 1 stood against the front wall and stop 3 at the door: the reverse of the unloading order. We did not test whether typing the stops in reverse puts them right. cbm3.net takes a stop per line, and its plan put a stop 1 carton in last, but it says “Strict stop order is off, so the delivery sequence is not enforced”, and the result was the same with it on. Gateway Lines has no stops. Ours never places a piece for a later stop between the door and a piece for an earlier one; on this load it gave each stop a lane along the length, with stop 1 reaching the door.

Load on top and weight. Only cbm3.net and ours take a limit on the mass that may rest on a piece; on the EUR pallets, containerloadcalculator.com stayed within the 500 kg limit without being told it. On the tile pallets, where the payload runs out before the floor, containerloadcalculator.com says “This load reaches the weight limit before the volume limit”, Gateway Lines leaves ten pallets out as “over weight”, and our planner shows the payload used, 99.1 % in the first container, and in a single container gives “the payload is used up” as the reason for what stays out. cbm3.net also stops at 20 pallets per 20 ft — with the same pallets at 100 kg it loads 22 — but its weight panel reads “Payload: not assessed”.

On density our planner is level with containerloadcalculator.com, not ahead of it: eight cartons more on one load, a tenth of a point and one piece less on two others, and the same count on five. Those are the loads ours was tuned on; on a load nobody tuned for, we would not expect a clear winner.

On speed, containerloadcalculator.com returned every plan in 0.6 to 1.7 s, page load included, in the browser, with no account. We gave our search 1 s per container and timed it outside a browser, so we make no claim to be faster. cbm3.net took 2.5 to 6.4 s, and Gateway Lines, drawing an animation, 7 to 46 s.

On inputs, cbm3.net takes the most per line of the three — this side up, not stackable, load on top and a stop for each line — and says aloud when stop order is not enforced. Gateway Lines names a reason for every piece it leaves out, which is the behaviour we copied. All three give a result in a browser in seconds without an account, and that is the bar.

And our planner refuses more than the others: cartons that would sit under the load line of a Hapag-Lloyd or Maersk reefer but over CMA CGM’s are left out everywhere, which costs volume in the boxes with the higher line.

Goodloading has no demo without an account: its demo address returned an error page, and the app opens on a login that offers the free version after sign-up. We did not create an account, so it has no results here. The SeaRates load calculator could not be opened on 7 or 8 October 2026, because searates.com and its subdomains had no address record in public DNS on either day. Both will be measured when they can be reached without an account.

Only free tiers were used, on one machine, in one browser, with each tool’s preset containers. No paid plan, mobile app or custom container was tried. Each load was entered once per tool; only our own planner was run twice. Where cbm3.net’s weight and containerloadcalculator.com’s second-container weight are blank, the tool showed none.

cbm3.net’s and Gateway Lines’ plans were judged from their 3D views and screenshots, not geometrically, so a piece hidden behind others could break a rule without our seeing it.

The ten lists — every size in millimetres, every mass in kilograms, the rules of each line — and every figure in the tables above are published as load-study.json, free to reuse with attribution. Anyone can type them into any calculator and compare.

If you build one of these tools and our reading of it is wrong, or it has changed since 8 October 2026, write to us and we will measure it again and publish the new result with its date. How we check figures is on how we verify data, and corrections are logged on the about page. The rules our own planner follows, and their sources, are on the load calculator page.