Posts from Skills for an Inferno hero.

I would suggest leveling up "Fundamentals of Leadership - gather troops - master of the Gate and also the speed aura" instead of Chaos magic; this will be a plus, especially for this hero. As for Heos, it doesn't make much sense because his main stats are attack and knowledge.

I would rather level up "Fundamentals of Leadership - gather troops - master of the Gate and also the speed aura" instead of Chaos magic; this will be a plus, especially for this hero. As for Heos, it doesn't make much sense because his main stats are attack and knowledge.

Thank you, I will try it! It's a pity that "Leadership" has only a 2% chance of appearing. 8% for "Fundamentals of Dark Magic" is also not very encouraging; I really want to guarantee getting "Call of Urgash." Thank you again!

Luck with wide gates, Leadership with the gatekeeper, Attack with battle frenzy, Logistics with the gatekeeper, and Dark Magic with all the summons – this is an unbeatable build for a demon. Chaos Magic will only be good on small maps, and even then, not always. When fighting against superior enemy forces, only Dark Magic is effective, no other options. And if you're also lucky with the Call of Urgash, you can wipe out everyone without any losses. Ballistics is a useless skill for a demon on large maps. Even if it's a Deleb. It might hit a stack a couple of times, but then a smart opponent will destroy it, and the entire skill slot goes to waste. Against legionary armies, ballistics and chaos magic are a pointless waste of time and effort. Defense is also useless for demons. If the enemy attacks, there will be significant losses anyway, no matter what. It's much more effective to simply prevent the enemy from attacking.

Luck with Gatekeeper, Leadership with Gatekeeper, Attack with War Frenzy, Logistics with Gatekeeper, and Dark Magic with all spells – this is an unbeatable build for a Demon. Chaos Magic will only be good on small maps, and even then, not always. When fighting against superior enemy forces, only Dark Magic, no other options. And if you're also lucky with the Call of Urgash spell, you can wipe out everyone without any losses. Ballistics is a useless skill for a Demon on large maps. Even if it's a Daleb. It might hit a stack a couple of times, but then a smart opponent will destroy it, and the entire skill slot is wasted. Against legion-sized armies, both Ballistics and Chaos Magic are a pointless waste of time and effort. Defense is also useless for Demons. If the enemy attacks, there will be heavy losses anyway, no matter what. It's much more effective to simply prevent the enemy from attacking.

I agree with everything, but:
Ballistics can be useful even on large maps – it depends on the availability of a mentor and money. If there are no money problems on the map and the mentor is nearby, Ballistics will make life easier in the early stages of the game. Since the opponent runs around with a small army, and it's difficult to destroy the Ballista's hit points + it will be effective against early-game mobs with its 3rd shot (especially against archers and mages with 7 Imps). It's not the best option in Multiplayer, because if you split into 4 stacks, you're guaranteed to lose, and that's a waste of money on ransom – although it might be worth it. After a month or two of playing, Ballistics will no longer be needed.

But on huge maps, education through mentorship can be useful. Simply to level up all additional heroes to around level 25. With Berserk or Subjugation, these heroes can harass the opponent well with minimal losses, but this is more for solo play.

luck, ini, defense, chaos, magic.
P.S.: I hope they bring back the demons and the Ifrit specialist for the weekly Grail, that would be fun!

I always take Attack, Logistics, Luck, and Dark Magic, but the last skill depends on the situation. Sometimes I choose Defense to compensate for the relatively low defense of Inferno units, and to deal damage with Hellfire when surrendering. Sometimes I choose Leadership to summon more units, especially if it's Nimus, and morale is also useful. Machines are also a good addition. However, Chaos Magic isn't very effective. Although demons are supposed to be proficient in this area, as creations of Urgash, they don't seem to excel in offensive magic, and it's not worth taking just for the single Hellfire enhancement. However, Light Magic, strangely enough, is also quite useful for demons.

It all depends on the tactics and the opponent.

Added after 32 seconds
Logistics and attack are the most important.

Added after 42 seconds
And then, choose based on tactics and the opponent.

Added after 1 minute
Heroes... they are all good.

And Gates are an integral part of Inferno.

Added after 3 minutes
By the way, against the League (and in general), Cursed Land is useful... let them run around as much as they want :smile40: :smile40:

Added after 11 minutes
Important and a priority... Hellfire + Withering Flame, Cold Steel won't hurt, Aura of Swiftness, Death Stride, and Path of War/Fast Gate Opening, Cursed Land and Mark of the Cursed + Weakening and Devil's Strike, Sulfur Rain or Rapid Fire Ballista (depending on the situation), Call of Urgash is very useful (in long and drawn-out wars on large maps)

I've been reading this forum for a long time, but I've never written anything. Inferno is my favorite faction in HoMM5, so even if it's a bit slow, I'll throw in a few thoughts. I've been playing them for a long time, and not just against the AI; I play Multiplayer (by the way, it was funny reading this thread — it's immediately obvious who is a practitioner and who is "just arguing for the sake of it," I won't mention nicknames).

Preamble over. Regarding skills.

1. I agree that there are 2 choices — rush (almost always siege engines via Delab) and late game (variation: super-late). I can't stand engines; Delab has always made me sick, so I'll write about the late game.
The skills I chose in the poll and consider priorities are Logistics, Attack, Luck, Defense, and Magic (usually Dark, possibly Light, or in the case of Grol and playing against dogs — Chaos).

For the late game (for me, level 20-23 in a real game), I consider the following build mandatory.

Expert Dimensional Door - Soul Eater.
Expert (Advanced) Logistics - Pathfinding - Fast Dimensional Door - (Surprise Attack).
Expert (Advanced) Attack - Battle Frenzy - Mass Haste - (Tactics).
Constant (High) Luck - Soldier's Luck - Wide Gates.
Expert (Advanced) Defense - Endurance - Mass Stone Skin - (Evasion).

Expert Dark Magic - Mind Control - (Pain Master).

This is realistically achievable by level 23; without Dark Magic and the perks in quotes, by level 18. The priority is specifically Logistics, plus Attack and Defense with Haste and Stone Skin, and Luck with Gates for a 100 percent chance of a gate.

Logistics is important not only for fast gating and occasionally Surprise Attack, but also for efficient map clearing (which means good artifacts and a couple of extra levels) and battlefield control (catch and kill — run and escape).

Attack is important not for its 15 percent damage, but for Battle Frenzy, which increases damage for the same devils by almost 1.5 times, plus the 30 percent initiative (the person who suggested taking more Luck instead of Haste apparently has no idea about DPS; by the way, moving more often means hitting more often).

Luck - double damage, a higher chance of fear from Nightmares, and most importantly — a chance to double the summon.

Defense - a very important skill for a demon, although some doubt it. For races with low base defense (Demons, TE), it saves more damage than it would for Dwarves, Knights, or Necromancers. Stone Skin was mentioned dismissively, but it provides a full 9 stats. In a real game without relic artifacts, a demon will have 5-6 defense at most. With Stone Skin, it's already 14-15, comparable to an unbuffed Knight.

Dark Magic - surprisingly, I take this as a filler because, in practice, parties often end before I have time to level it up and build guilds. Undeniably a weighty skill, but only after the first four.

In super-late game (levels 25+), which is usually played with a friend or munchkins, it's realistic to max out these skills.

This is practically the ideal build. A demon with these perks is fast, deadly, and can take a hit. Just try playing this way, and you won't need any engines. Grol fits here perfectly, but isn't as indispensable as the notorious Delab.

Now a few words about the other skills I find debatable — engines, Education, Sorcery, and especially Leadership (I won't consider magic, I'll leave the eternal Dark vs Chaos debate aside; most would agree that Dark is preferable for a long game).

Engines - the skill is good for a short game, but useless in a long one due to quite low damage against crowds of meat and the enemy's ability to wipe out 1100 ballista HP in one hit/cast. Distracting the enemy for one hit at the cost of 5 level-ups isn't worth it. I've never taken it and don't take it due to a purely personal dislike for engine-users (I consider it a bit of a cheat to pass neutrals so easily), so I admit I'm biased.

Sorcery - it was already mentioned above that a demon is primarily a warrior. Sorcery just wastes a slot, as it doesn't affect mass spells, and taking it in hopes of casting level 4-5 spells more often is wasteful. A demon's fight should end sooner. There are no useful perks except Wisdom for the Phantom. Distraction costs a whole 6 level-ups.

Education - someone wrote here that it's foolish to neglect Education in a long game. I disagree, and here's why.
The demon is a power-based unit, so one should use Education to increase Attack or Defense (we aren't betting on Spell Power, and Knowledge is already high enough + devils will steal mana).
But of all the stats from Education, 30 percent will go into Knowledge and 15 into Spell Power. Every second stat will go into Attack. At level 20, the bonus from Education = 10. 4-5 will go to Attack, 1 to Defense. Sometimes it's even worse. Taking the skill just for the basic level of Punishing Strike is a questionable pleasure.
The funniest thing is that among power-based units, Education is good for Knights (75 percent into combat stats) and ideal for Barbarians (85 percent). I often take it for them when possible.

Leadership - I simply don't understand this one. Taking it for itself is wasteful — army morale is already high in the game; for heroes with 0 morale, creature morale = 2, why do you need another 3? Regarding perks — Empathy and Inspiration from 2-percent Leadership go through questionable 2-percent Education, so forget it. Aura of Speed is not bad, but there is Tactics (and speed boots).
Gate Master - loses miserably to Wide Gates of Hell. The latter double the gate with a decent chance from 50 to 100 percent, while the former only increases it by 20 (though stably).
Of course, one could argue that there is the Sorrow spell, Necromancers with Banshee screams, etc. But I still believe that by the finale, positive morale will be far beyond 3-4.

Something like that.

Everything said above, despite some absolute tone, is the author's heavy IMHO. Good luck to everyone and wide gates (if anyone is still reading this forum). ^_&

I largely agree with the above post. Logistics, attack, defense, and luck are always my priorities for this faction. The same applies to perks, but for luck, it's desirable to also take resistance, especially if you expect Darkness or Chaos from your opponent. I also don't like machines for Demons, even though ballistae have the highest damage for this faction. They are unreliable, especially knowing that machine operators are easily countered by one perk in Chaos. Besides, they are only useful at the beginning, and then luck will provide much more benefit. As for magic, I tend to favor Light. Mass Curse and Haste for a faction with high initiative and retaliation will be a big plus. Bless is also good, as Demons have the largest damage range, which is very useful. And Mass Stone Skin and Mass Dodge are also very good, as Demons often struggle against archers. But since Light is difficult to develop, the choice often falls on Darkness. I wouldn't be so categorical about Leadership. Morale can, of course, be obtained through artifacts, but the speed bonus for Demons is very useful, as the opponent may reduce their speed with Necromancer artifacts. Especially Archdemons lack it. Without a speed bonus, you have to stand in direct line of fire to take out archers. Tactics can also be countered if you're not playing as Nibros. But luck is still more useful. Sorcery is clearly not the way to go. However, the developers managed to give it to two heroes at the start, even Alastor, for whom it is completely useless. I don't understand this logic. At most, it can be taken for Chaos, but I rarely level up this school for Demons for one well-known reason. Education is still good for any faction in the late game, although the perks there are not very useful for Demons. So, in general: Logistics Attack Defense Luck Light or Darkness

For Magistar: I take Resistance, and often, I almost forgot to finish writing about it. I was writing the post late at night. Regarding Light magic – similarly, all other things being equal, I will always choose Light. Firstly, to compensate for the damage "gap" (of course, I'll also use spells like "Skin" and "Burn"). Secondly, to surprise the opponent. I was simply writing according to the standard, a "classic" build, and since Light relies on a lack of spells, I chose Darkness (in my post, Light is listed as the second choice). I won't argue about Leadership; I meant it in comparison to Luck. I haven't encountered situations where horses and cavalry units couldn't reach the enemy line without a speed aura. The developers somehow managed to give Sorcery not so much to "even Alastor" as to Jezebel, a specialist in succubi. This is a complete failure. Why would archers need Sorcery? Dugald has ranged attacks. In this regard, the Inferno heroes are the worst – Grol (Chaos), Alastor, and Jezebel (Sorcery) have been ruined by their perks, plus Daleb (banned because of machines). You can only play with half of them. In other races, 1-2 heroes are ruined by their perks, and even that is debatable. After all, the Inferno faction is not liked in all parts of the game. The recent news about the failure of Heroes 7 and, consequently, the add-on with the Underworld confirms this.

I won't argue about leadership; I meant it in comparison to luck. I never had situations where horses and cerberi didn't reach the enemy line without a speed aura.

Luck is better; there's no arguing about that.

The developers somehow managed to stuff magic not so much for "even Alastor" as for Jezebeth, a specialist in succubi. This is a complete failure. Why do archers need magic? Dugald has ranged attacks.

It's a failure in both cases. For Jezebeth, this magic was also neither fish nor fowl in the scenario; it only spoiled the entire build and speed of completion.
Alastor is a shadow creature; there's no point in spending another branch on magic, as it doesn't affect mass battles anyway.
Just for example. Nazir, a specialist in fire, starts with chaos. Svey, a specialist in lightning, starts with chaos. Faiz, a specialist in beams, starts with darkness. Vlad, a specialist in raising the dead, starts with summoning. Even the lords have corresponding skills. Alastor...

In this regard, the Inferno heroes are the worst – their perks are messed up for Grol (Chaos), Alastor, and Jezebeth (magic), plus Daleb (banned because of machines). You can only play with half of them. For the other factions, 1-2 heroes have messed up perks, and even that is debatable.

Yes, in the end, only 4 heroes start with decent perks. Although, does Marbas really need such a strong "reflection" if he relies on resistance? But starting with defense is still good. Well, Daleb can also be counted; after all, she's a specialist in machines, and if you play as her, you play through machines.

Inferno heroes are disliked in all parts. The recent news about the failure of Heroes 7 and, accordingly, the add-on with the Underworld confirms this.

Even the developers are obsessed with this unimaginable banality of "Good is stronger than evil" and other nonsense. Bias should not exist at all.

Even the developers are obsessed with this unimaginably banal "good is stronger than evil" and other nonsense. There shouldn't be any bias at all.

+1. In both the third and fifth games, the demons aren't that great. In the fifth, they're much more interesting, but still... there's too much hassle with breaking through defenses, and the hero selection is not ideal.

I read your level 40 Nimue build, I was just searching for how much bonus the specialization gives, and I came across your post. It was informative and helpful, thank you.)

Regarding the build - a small nuance about the defensive perk. If you're going for light magic, this perk is definitely the way to go; it will make breaking through defenses much easier, and if you're focusing on dark magic and summoning, extra base survivability won't hurt. I had a battle against Nimue for Axel on the 3rd month, and the demon only won because of mana and the ability to control paladins.

I want to test one strategy against aggressive factions, specifically elves, who demons have trouble with, although it can be tried against anyone.

The idea is to take logistics and keen mind, light magic with stone skin and storm shield, standard attack, luck, and either a defensive branch with defensive posture or even stubborn defense or readiness (a very debatable perk for a demon, but still).
You'll have to sacrifice surprise attack (not a big deal) and either pathfinding or fast gate opening (in my opinion, the former, but it needs to be tested).

An important point - artifacts should not be chosen solely for their defensive properties.
Ideally, you'd want the dwarf set, which gives stone skin and dodge on expert, but if that's not available, then:

- Dragon's Tongue and Dragon Ring - together 4 defense (if there's no Inferno Staff), if not, then "On the Edge of Balance":
- Dwarf Shield and Cuirass - together 8 defense and 40 percent resistance;
- Maybe Sandals of the Saint and Secret Protective Cloak, although there are better options - together 2 defense. In total, 12-14 defense.

In upgrade locations like arenas, don't choose attack, choose defense. Many people (including myself) automatically choose attack and end up with a berserker with stats like 35-5-10-15. But we need to try to create something like a dwarf, something like 20-20-10-15, all numbers are approximate.

Battle algorithm.

First, cast stone skin on the hero with keen mind (if you don't have the set) and summon, then stand in defensive posture and buff the troops. Of course, not all the time, only until the gate is summoned. Then attack with the entire force.
The idea is that the troops should withstand the initial attack due to their starting defense and the applied skin, and retaliatory strike will compensate for the lack of attack.
If summoning isn't an option (you only have time to open the gate and then get hit by everyone), immediately attack with the troops under stone skin. It's hard to say for sure, as initiative varies depending on the artifacts, and there's also a 25 percent variance.

You can also try it without light magic, with the mass skin perk in the defensive branch, or you can try using keen mind to cast mass slow or immediately bless-retaliatory strike; variations are possible.

It might turn out to be nonsense. Or reinventing the wheel, as the game is quite old. But maybe someone will find the idea interesting. Or maybe someone has tried it, let me know the results.

P.S. I think taking defensive stats for a demon is the optimal option, in any case. Initiative and good damage, combined with survivability, should give good results.

+1. Both the third and fifth factions have demons that aren't very good. The fifth faction is much more interesting, but still... there's too much hassle with recruitment and the hero selection is not ideal.

Overall, I like how they are designed, including the racial ability. But it's a difficult faction, especially for beginners, which is why many people are turned off. Recruitment is somewhat difficult, but it mainly affects archers and mages; dealing with melee units is generally easier for some factions. However, with the right tactics and upgrading the dogs, you can compete with archers. I find it harder to start with the Dwarves. But in general, there are some drawbacks that could be improved.
And the heroes are not impressive. For some reason, there are few heroes specializing in creatures, only dogs and succubi, which is clearly not the best choice. But heroes for imps, demons, and horses would be better. And the starting skills of many heroes are especially appalling, especially two useless spells.

I read your test of a level 40 Nimue, I was just searching for how much the special ability gives, and I came across your post. It's informative and insightful, thank you.

I haven't fully figured it out myself yet, and that's because of the strange way the set works, which gives percentages to everything except the racial ability itself, although it somehow takes it into account when a gate is successfully opened.
And if you're interested in this topic, I can recommend a good video on how to open gates:
https://www.youtube.com/watch?v=V2Ifu9ZEfzA

Regarding the build - a small nuance about defense. If you take light, definitely take this perk; it will be much easier to break through, and if you focus on dark magic and gates, extra base defense won't hurt. I had a battle with Nimue for Axel on the 3rd month, and the demon only won because of mana and subduing paladins.

I want to test one strategy against aggressive factions, specifically elves, with whom demons have problems, although you can try it with anyone.

The idea is to take logistics and keen mind, light magic with stone skin and storm, attack as standard, luck, and either a defense branch with defense or even with stalwart defense or readiness (a very debatable perk for a demon, but still).
You'll have to sacrifice surprise attack (not a big deal) and either pathfinding or fast gate opening (in my opinion, the first one, needs to be tested).

An important point - artifacts are not taken last for defense.
That is, ideally, the gnome set, which gives stone skin and dodge on expert, if it's not available, then:

- dragon's fiery tongue and dragon ring - together 4 defense (if there is no staff of the underworld), if not, then "on the verge of balance":
- shield and cuirass of the gnome set - together 8 defense and 40 percent resistance;
- maybe sandals of the saint and secret protective cloaks, although there are better options - together 2 defense.
In total, 12-14 defense.

In upgrades like arenas, don't take attack, but defense. Many (including me) automatically take attack and end up with a kind of berserker with stats of 35-5-10-15. But we need to try to make something like a gnome, something like 20-20-10-15, all numbers are approximate.

Battle algorithm.

First, cast stone skin with the hero with keen mind (if there is no set) and open a gate, then stand in stalwart defense and buff the troops. Of course, not all the time, only until the gate opens. Then attack with the whole mass.
The point is that the troops should withstand the blow due to the starting defense and the applied skin, and the devastating strike will compensate for the lack of attack.
If opening a gate is not an option (we only have time to open the gate and then get hit by everyone), immediately attack with troops under skin. It's impossible to say for sure here, as the initiative is different depending on the artifacts, and there is also a 25 percent variance.

You can also try without light, with the mass skin perk in the defense branch, you can try to cast mass slow or immediately bless-devastating strike with keen mind, variations are possible.

It may turn out to be nonsense. Or reinventing the wheel, as the game is quite old. But maybe someone will find the idea interesting. Or maybe someone has tried it, share the results.

P.S. I think taking defense stats for a demon is the optimal option, in any case. Initiative and good damage, along with survivability, should give good results.

Yes, I've always considered light to be a top-tier magic for demons. Defense is also important, which is true, and you need to protect the imps. That's why it's problematic for races like humans, as they often run out of mana. With such a setup, darkness also plays well, as there is practically nothing to fight against high darkness without mana.
In general, there are many tactics, you need to look at the situation. And too much in Heroes is decided by randomness. I haven't played much against humans lately, I don't have time, except for a game on impossible difficulty just to test the AI for all races. I also consider the AI for demons to be one of the best, although the path to it is not impressive.

I played a demon vs. elf matchup against a skilled elf (attack-retribution-leadership-haste aura - defense - light-luck), and I've confirmed the complete uselessness of readiness and gating upgrades for demons, even though the elf wasn't Wing, but Ossir. Since the elf also happened to be my best friend), we replayed the final battle several times with various upgrade variations. I played as a mentored Grol, two specialists in level 3 creatures, although those same same own "cheater" blue shooters didn't want to be countered by dogs.)
This matchup prompted me to test gating and upgrades against all Factions, which I did for some time, while simultaneously figuring out how gating works in percentages (I'll post the table below). During the tests, I reconsidered the usefulness of several skills.
By the way, I realized why I was so categorical in my opinion. The thing is, I used to play the fifth games less than the browser HWM, as many probably know. There, a demon could easily have 25-30 defense because the player distributed stats themselves. Therefore, you could calmly wait for the gate. Plus, since the player didn't know which opponent would appear before the battle, everyone tried to pick a universal build. It doesn't work that way in the fifths. Now I play people via tangle; I've given up on the browser version. In summary, my tests and matches with people allowed me to think through and reconsider a lot.
Regarding skills.

1. War Machines.
Probably every demon player eventually realizes that in short games, you can't do without them—the lack of low-level shooters and the fragility of the army forces you to take it. I don't play rush maps, and in long ones, I retrain via a mentor in extreme cases, so I still try not to take it. But if you must, the sooner the better: first the tent, then the ballista—minimizing losses and additional damage = faster leveling.

2. Defense.
I remain of the opinion that it's a must-have for demons. In earlier posts, I forgot to mention "Fight to the Last," or FTL. This perk is very necessary specifically for demons, as it not only helps clear neutrals with single units but also allows the gate to work longer in the final battle. In short, I've concluded that if the match isn't prolonged—5-6 weeks—and/or the opponent isn't high-damage, I'll take Mass Skin. If the match drags on or the opponent is a one-shotter (Light Elf, for example), FTL is better.

3. Education.
In tests, I noticed an interesting point—to my shame, I didn't know that if you reset Education and take it again, stats are distributed differently. So once I managed to get 9 in attack and 1 in defense out of 14 stats, which is significant. In short, if you have the money and a mentor—experiment.

4. Leadership.
I underestimated it. In battle against Necros, for obvious reasons, it's a must-have, and Gate Master works better than Wide (I realized this during gating tests) = it gives a stable 20 percent bonus unlike the 40 percent to only 1-2 stacks out of 7.

By Factions. I'll try to describe briefly, from "worst" opponents to "best" (here my opinion coincides with the thoughts of the respected DemonSD, who almost single-handedly defended the interests of Hell on the AG.ru forums). Assuming a non-rush game, roughly 6-10 weeks. Attack, defense, and logistics are almost mandatory in all pairs; the fourth skill is magic, the fifth—if possible.)

1. Elves.
In my opinion, the racial skill doesn't need to be leveled up; there won't be time to gate. In defense—FTL, but the skill itself seems debatable because of elven one-shots (since there will be no gate, there will be no one to deal damage while the base units are alive, unless you hope for darkness). Light with Storm Wind is mandatory, even without spells—I think everyone knows what emerald-crystal dragons and nymphs with haste aura and retribution are capable of.
Keen Mind is desirable. You can place fire hounds on the 3rd cell from the right and left edges and pray to Urgash that the shooters stand in that corner. Then as the first custom, we set Punishing Strike, and the dogs crash into the crowd of enemies, hitting the shooters with breath. As a fifth skill, I would take darkness (the only option with 2 magics)—hoping for mass confusion, berserk, or with good spellcasting—subjugation. Overall, a very difficult opponent; I'd say the anti-faction of demons.
Furthermore, I realized that my advice to take artifacts and defense boosters was wrong. In this battle, it's better for a demon to have stats 30-5 than 20-15, because due to those cursed armor units, they still can't hold a hit, but deal less damage.

2. Academics.
Mages are very versatile; the only thing that can be said is—you can't drag it out too long, giving them the chance to pump out mini-artifacts on initiative, but you also can't rush hoping for machines (pests). Don't abandon single berserk-subjugations (mirror). Gating is optional—if there are no minis on initiative, then yes; if there are, it won't help anyway.)

3. Barbarians.
Orcs close out the triad of the most dangerous opponents for demons, and they are the ones I haven't figured out what to do with yet. In tests, a demon army with 40 attack (31 + 9 from education) under light crashed into a barbarian army, managing to wipe it before rage built up; in real games, you rarely get lucky with such stats. The orc himself doesn't let you wait for the gate or darkness, while simultaneously blocking darkness with goblin shamans. Everything is optional; you can try both light and darkness. In darkness, blindness is better than berserk, as an orc stack under berserk gains rage twice as fast. Stacks must be taken out one by one, but keep an eye on the rage of others. Cursed Ground, plague, and unholy word are good at reducing rage.

4. Dwarves (in late game).
Charge immediately after gating (fast opening is mandatory). Light and darkness are equal in my opinion—darkness looks slightly better, but it's countered more by dwarven luck and the rune of exorcism. Resistances don't work on buffs, but there is a magic oversight rune. Stacks should also be taken out one by one (resurrection rune). Storm Wind performs well, because although there is only 1 flying creature opposite, it's worth a whole dwarf army.) Overall, hope for faster development due to logistics (same with mages).

5. Warlocks.
Another race, along with elves, where gating doesn't work; you must attack immediately. Unholy word and especially cursed ground perform well; against tricksters with stalkers = darkness is almost a must-have. You can try to level chaos (or take Grol) for Armageddon. And here you can actually level machines if the game moves to small growth (though with high SP, TE might blow up the ballista in one cast). Guard imps like the apple of your eye; warlocks fear them most of all. Overall, against a standard warlock tuned for destructive magic, the chances are good—you just have to prevent tricksters.

6. Necros.
In the late game, if they aren't cheating with a death herald-necro-converter in the castle (saw a replay with 900 vamps under Lucretia around week 8-9), not too difficult an opponent. The only race where you can skip fast gate opening, and the only one where you can fully gate (and even without it). Light with storm performs very well (the necro has 3 flyers; summons will be crippled with 3 speed) and leadership, for obvious reasons. I don't take FTL because there are often situations where my single unit (a nightmare, say) blocks my own path to a vampire or zombie in the corner. Against berserk, sudden attack helps a lot—just teleport the berserking stack to the necro. In general, everything is fine here in a medium-to-long game.

7. Knights.
Despite hordes of paladins and generally recognized cheating in long games, not too difficult an opponent (if you don't let it get to those hordes), or maybe I just ran into the wrong players.) The key to victory, as with warlocks, is imps. Knights have poor knowledge growth; accordingly, without magic, they can only hope for prayer-inspiration. Darkness is mandatory here—in my opinion, there's no point hoping to out-hit knight troops with excellent defense-attack stats using your own damaging but fragile ones, even under light. Berserk, confusion, blindness, subjugation—look much more effective. Destroy them again stack by stack, starting with the high-initiative ones (griffins).
And in general, the same principles of fighting enemy stacks apply—using initiative and having 14 stacks against 7, swarm one with mass, then move to the next, as demon stacks are a bit fragile in 1v1 combat.

Roughly like that. As you can see, there is no universal recipe for victory; everything is learned through comparison.

Heroes should be discussed in another thread, but briefly I'll say again—in long games, creature specialists perform well, especially tank specialists (Laslo, Tilsek, Ingvar, Orson). Demons have problems with both tanks and existing specialists.
There was one person on the Nival forum, nickname Sta1kerrr, I think. He was the first to notice that if demons had a level 2 specialist, he would play him. People almost laughed at him, but as far as I'm concerned, he's right. Having a fat stack of OD or jumpers with +10-12 defense from the hero, and under skin, the demon would be much more durable.
Regarding existing ones—I tested Grol and Jezebeth. For myself, I came to the definite conclusion that Grol should not be used except in rush games, even with a mentor. Dogs can't hold a hit; all they can do is bite hard once, but whether that hit is worth choosing Grol is a big question, especially since sometimes they aren't even allowed one hit.
Jezebeth looks more promising—after all, 1.875 damage from demons to 4 targets, and I read that with luck the modifier more than doubles (1 stack - full damage as if direct, 2 - full damage as if direct, 3 - half, 4 - quarter, info from DemonSD), and under Jezebeth, the base attack of 6 for demons becomes 16-18. Additionally, Jezebeth increases demon defense; from a base of 6, it again becomes 16-18. At level 23, the parameters of demonesses under Jezebeth equal those of nightmares.
But still, Jezebeth cannot be a universal choice like Tilsek-Ingvar. I checked: against elves and humans, taking her is pointless—elven blues even without Ossir wipe out all demonesses in one shot; with humans, 290 shooters under knights, having lost only 60 to demonesses (damage showed that even with punishing strike under keen mind it was only 43-86), wiped out 100 demonesses out of 101, leaving a single unit with 1 hp (FTL, yes).
So, elves and humans are out. Jezebeth performed well against dwarves—wiped more than half the elders (26 out of 49 I think) in one shot, and against orcs. I'm sure she'll be good against necros and dark elves. Mages are optional (if they let you shoot).
But for now, I'm trying to use Jezebeth even without a mentor—wisdom sometimes gives a phantom, and distraction helps counter casters well. One problem is surviving to a high level, as sorcery doesn't help in breakthroughs, and there's no skill synergy (for example, Nibros at level 3 has attack-tactics-BB, while Jez at best has base charms-wisdom, 1 attack-tactics or BB).

I haven't fully figured it out myself yet, and that's because of the strange way the set works, which gives percentage bonuses to everything except the racial skill itself, although it somehow takes it into account when a gate is successfully opened.
But in general, if you're interested in this topic, I can recommend a good video about opening gates:
https://www.youtube.com/watch?v=V2Ifu9ZEfzA

I haven't watched the video yet, but here's the table I promised, with some comments.
The table is in a notepad format because it looks awkward if you just copy the numbers into the post. The numbers mean so many creatures out of a thousand, i.e., 525 means 525 out of 1000 summoned, or 52.5 percent, which is more accurate.

Conclusions.

1. I've figured out for myself that Wide Gates of Hell only double the percentage of the upgraded racial skill, without additional bonuses from castles or artifacts.
That is, with a standard 1 castle and skillful gating, it will be (40*2)+10 = 90 percent, and not (40+10)*2 = 100. Similarly, with 4 castles, for example, it will be (40*2)+40 = 120, and not (40+40)*2 = 160.
Thus, the static bonus from the Gate Lord in Leadership looks more convincing; all stacks will guarantee summoning 40+10+20 = 70 percent, which is better than 5 stacks of 50, 2 stacks of 90 percent.
The amulet works the same way, i.e., it will be (45*2)+10 = 100 percent.

2. Nimue's specialization adds a percentage per level, so at level 20, he is as if he has a Gate Lord or 3 castles instead of 1. It's a good bonus, but only against half of the factions. I haven't decided whether to give him Wide Gates or Gate Lord; building 2 branches is risky – defense is needed. Here, perhaps, as an exception, Wide Gates, because it will guarantee 70 for everyone, plus a chance of 110 percent for 1-3 stacks. Moreover, luck is built in.

3. It was once again confirmed that the developers have a dismissive attitude towards demons. I'm talking about the "Will of Urgash" set. The description says that it adds 25 percent. Logically, it should be 40+10+25 = 75. In reality, we have 40+10+2.5 = 52.5 percent. And then it linearly adds 2.5 percent, i.e., even with 6 castles, the bonus will be 15 percent, and it will only reach its promised effect of 25 percent with 10 castles, which will never happen.
There are other nuances in other variations (I'm not a mathematician; I can't derive an exact formula), but in any case, the percentage is negligible.
It's doubly frustrating, considering that the developers' favorites, the Dark Elves, have a set that works as it should, with a strange name, "Power of Runes," which quite nicely doubles the damage from chains, and no one is screaming about imbalance.
After testing, I put the demon set in last-to-last place (I don't like that the barbarian set reduces initiative); there's more benefit from 5 attack than from gating.

4. Considering the fact that all the perks for gating efficiency (except for quick gating) are in the second tier (after gathering troops, soldier's luck), I'm less and less striving for them and taking them. I limit myself to soldier and magesight, which is more useful. And in leadership, I often take the speed aura first.

That's it, it might be useful for beginners and those interested in the hellish army.