I played some games today and was impressed. There is still a lot to do and polish, so I won't comment so much on balance or minor visual things. A lot of the effects and models seem to be placeholders or WIP.
I also haven't really compared these units head-to-head with those of other factories or really looked into the stats. This is hampered by the lack of a repository to refer to (I could extract the sdz, but that can wait). So don't go violating Quant's Rule just because I say things like "unit X might be a bit slow/tough/whatever".
Ranger - This feels like a reasonable design. Cost is between Conch and Welder but it is fairly fragile. The fire arc was very annoying but I hear you've since fixed this. It felt fairly slow. My main concern is that it doesn't fit as a bot con because it's name doesn't start with con.
Ares - As the light raider I have the least to say about this until it is really tested head-to-head. As it is, Bandit speed makes it pretty close to Bandit. I am a little concerned by Ares vs. Ares always resulting in one survivor on 1% HP. In other raider mirror matches the projectiles flying through the air create a lot of space for equal trades.
Cougar - This initially felt like it made Ares obsolete, but with more play I saw that Ares had a small niche. Cougar is definitely not a riot due to its speed and riot matchups. Ares and Cougar should just be the two raiders. I was most impressed by how bad Cougar seemed to be against defenses, and at dealing sustained damage in general, which gives Ares a niche. Ares could lean into that niche with more speed and/or damage. It is fine for a factory to have a main raider and a more niche one, and these two seemed to share space a bit more evenly than most. I would expect Cougar to end up with nerfs when compared to other factories.
Catapult - This unit is ok. I barely built it due to what was going on in the rest of the factory, but short-ranged burst AoE artillery is a decent role to fill. The high AoE risks inducing a lot of Fighting the UI in the enemy, because the shots really shouldn't be hitting clumps of units, yet they are devastating when they do. The downside of high trajectory and inaccuracy is that most turrets are small when viewed from above. This is the problem Badger mines were designed to solve, but the high AoE of Catapult also solves it. Whether it creates too many other issues remains to be seen.
Apollo - Didn't really build this either. 4.2k health for 630 is awkward and less than Minotaur. The projectiles also seemed very fast for an assault so it is hard to tell whether things should be dodging them. They aren't plasma as the description suggests (they seem to be a speedy Halberd bullet), but visuals can be changed, although in general I prefer it if weapons are reused with similar visuals and stats, because it makes everything a bit more cohesive. A long range targeted manual fire on such a cheap unit is a lot, not even demi-striders get these. See Cold Take #39. It is also unclear why a bot factory wants a smooth weapon.
Iron Dome - Firstly, way back last decade, we decided names would be limited to one word. Plus might it get a bit dated? Anyway... this seems to be a heavy Angler and I'd be a bit worried about stepping on its toes. It should have script reload so as to not be very frustrating to use, or it could have individually larger missiles. I would not want to make it cost any more than 500 as AA shouldn't be survivable. That would ruin the ground interactions (so there is no AA strider).
Shiva - Shiva is a terrifying design. People were calling it "riot artillery" because it hits with the force of three Rippers at Stinger range. The AoE means it can outrange Stinger, which makes Catapult a bit obsolete. I'm not sure whether this design fundamentally breaks ZK or not. Units with more than around 420 range pay a premium for the privilege, often by having terrible stats or having trouble hitting anything (or both). So the question is whether Shiva has terrible enough stats. It has very slow turret turn speed, which may be a bit of a Fighting the UI issue in itself, but it does keep it from being amazing AA. The reload time of 12s is fairly long, although 1600 HP is quite a bit, and puts it out of Phantom range. I tried forcing Shiva for a few games, with some success. I suppose that once you have five of them you could have had a Dante. But I'm still concerned with how fun it is to play against. I expect Siegebot matchups to be defined by whether the factory can rush, tank, or outrange Shiva. That said, it wouldn't be interesting if the factory didn't have at least one unit on the edge of completely breaking the game.
Atlas - Didn't see this one in action. It's a bit weird that it has rockets and gauss, does it need this? Cyclops has two weapons, but it needs them because they do very different things. Perhaps this is ok at 2.5k metal. I am not sure how I feel like having a demi-strider that makes it sound like someone built a Detriment. The gauss spam bug seems to be easy to cause with this unit (I saw it after a bit of testing), which is ok for Det but probably not for such a cheap unit.
The Status Effects Unit - You didn't list this unit and I forget what it was called. The EMP:Damage ratio on the lightning is bad enough that it should have similar graphics and sound of Venom, not Knight. The effect for the slow bomb also seemed to be broken, so it was hard to tell what it was meant to be doing. I'm not sure how I feel about it. Two weapons on a 500 cost unit is a lot. They sort of do the same thing, but also one has AoE while the other is point target. But they are both status effects. That said, mixing status effects on one unit like this has a kitchen sink/mary sue feel. You'd probably get a stronger design if you picked, say, the lightning, and stuck with it.
Quantum Bomb Carrier - This is a simple idea that has been knocking around for a while. A lot of design work would need doing to solve all the obvious issues with it, where it would just delete large swathes of strategy space once the game reaches that point. Also, demi-strider cost prettymuch caps out at 2.5k. Even above 2k is a bit much. So if this unit were added, it would go in the Strider hub. The factory should probably have a bomb (just like the constructor should start Con-), in which case the obvious gap is disarm. Seismic and ground fire are other possibilities, although seismic would be a bit niche and weird for a bot, while fire maybe means the factory wants fire elsewhere too.
On the models in general I would suggest finding some artistic way to bring the arms in. Firing weapons from the extreme edges of unit is cursed for many reasons. Units can still look like they fire from the edges without having that many problems. Eg Reaver is mostly fine. The models that do this also look a bit similar when zoomed out.
I'm not sure what the theme of the factory is. Bot factories tend to be fairly theme, this is just heavy. Which are the units that the factory wants to build (ie because it follows its best game plan), and which are those that it builds to react to problems? These are the sorts of things that will take quite a while to figure out as the units are brought in line with each other. Names are also hard. Other factories have an overarching theme, or a few interacting mini-themes. We don't have that many names of gods, but iirc they're scattered over miscellaneous units and structures rather than being in factories. Siege equipment could be a mini-theme, which you've already got going with Catapult (perhaps we could put Ram in somewhere).