Bottom Line: Gunfire Games built one of the most mechanically generous action-RPGs of the decade and then hid half of it behind procedural dice rolls and a wiki tab. The combat and build depth are genuinely elite; the onboarding and optimization are not.
The gunplay is the thesis
Most soulslikes that add guns treat the gun as a ranged melee weapon — a slow, committed animation with a stamina cost. Remnant II doesn't. The shooting is fast, precise, and reads like a competent third-person shooter first, with headshot feedback, weapon-specific recoil personalities and a reload cadence you'll internalize within an hour. Then it takes the stamina bar, the dodge i-frames and the punitive checkpoint structure from the other genre and asks you to run both systems at once.
That combination is the whole game, and it works because the enemy design respects it. Root infantry pressure you at mid-range while something large and unpleasant closes the gap. You cannot kite forever — stamina says no. You cannot stand and shoot — the melee threat says no. The correct play is a constant negotiation between the two, and when a fight clicks, the rhythm is unlike anything else in the space.
The bosses are the best argument for the game. Gunfire's encounter design here is genuinely first-rate: multi-phase fights with add-management layers, arena hazards that matter, and — critically — bosses with alternate kill conditions. Solve the puzzle instead of out-damaging the health bar and you get a different reward. It's an elegant way to make a boss feel like a system rather than a wall, and more games should steal it.
The build system is the reason you stay
The dual-Archetype structure is the smartest thing in Remnant II, and it's smart for a boring, structural reason: it makes the second class a modifier rather than a compromise. Medic/Challenger isn't "half a healer and half a tank" — it's a specific playstyle with its own damage floor and survivability ceiling. Engineer/Gunslinger turns turret uptime into a DPS multiplier. Handler/Summoner is a legitimate pet army.
Because the prime perk of your secondary carries over even when you swap, you're incentivized to level everything, and because trait points, rings and amulets are shared across the account, experimentation costs time rather than progress. This is the correct incentive design for a loot game, and it's why the endgame holds. The Prism system and Apocalypse-tier Adventure runs give the theorycrafters somewhere to put all of it.
Then there's the part where nobody tells you anything
Here's where the review turns. Remnant II has no map. No quest markers. No meaningful in-game guidance about its own systems. In a hand-authored world, that's a defensible aesthetic choice — see: every FromSoftware game since 2011. In a procedurally assembled world, it's a structural problem, because the community can't build shared cartographic knowledge the way Souls players do. Nobody can tell you where the thing is. They can only tell you the thing exists, somewhere, on a roll you may not have gotten.
The result is that Remnant II's most celebrated content — secret Archetypes, alternate boss kills, hidden questlines — is functionally gated behind an out-of-game wiki and a willingness to re-roll worlds repeatedly. The game's own discovery loop doesn't surface it. I don't think this is a difficulty choice; I think it's a UX gap being sold as mystique. A single unlockable in-world map, or an NPC who hints at what's present in your current roll, would cost the game nothing in atmosphere and fix most of it.
The procedural generation cuts both ways for the same reason. It's the engine of the game's replay value and the source of its most common complaint: some players simply never see certain bosses or quest chains without deliberate, repeated re-rolling. Variety you can't seek is just variance.
The co-op flaw that should have been fixed at launch
Campaign progression in co-op only credits the host. Join a friend's game, invest twelve hours, and your own campaign hasn't moved. You keep gear, materials and trait points — which softens it — but the narrative and world-state progress belongs to whoever pressed host. For a game whose entire identity is three-player co-op, this is a design decision that has aged badly and still generates friction in every friend group that picks it up. Plan around it: rotate hosts, or accept that one player is playing and two are farming.
