Arknights: Endfield tier list — the ranking, and what each placement is actually based on

By DayOneFix Editorial · August 24, 2026

The tier list

Every placement below is what the two communities said, with who said it — we are not scoring operators ourselves. Attention figures are Bilibili play counts and YouTube view counts as cited in the sections further down.

Tier	Operators	What the placement rests on
T0	LaevatainZhuang Fangyi	Why — Laevatain is 「輪椅一姐」 (wheelchair queen) with the largest single-operator audience in the game — 叫我棉被’s guide at ~2.6M plays — and loops her ultimate. Zhuang Fangyi carries the game’s boldest single-character claim, 「T0劍仙」 (~1.4M plays)
T1	ArcaneLast RiteGilbertaMi Fu	Why — Arcane 「萬能術師，一訣雙用」 — one pull covers two jobs (~1.53M). Last Rite 「對單之王」 in CN and “Nuke Queen” in EN, arrived at independently. Gilberta 「什麼隊都能進的萬用插件」 (~990K). Mi Fu holds the EN Physical crown after the Pogranichnik rotation was walked back
T2	EmberCamilleLiinoRossiArdeliaEndministrator	Why — Ember 「余燼風評逆轉！榮登物理隊的神」 (~490K) — a Defender that plays like a carry. Camille 「三相之力，火隊必練」 (~660K). Liino 「掛機流輔助」 (~800K). Rossi enabled the first 「全新混傷隊」 (~590K). Ardelia is 「萌新之友」 and free. Endministrator 「管理員就是大C」 (~880K) — the free unit is a real carry
T3	PogranichnikLifengTangtangYvonneTyphoeus	Why — not weak — conditional. Pogranichnik’s armour-break rotation was called Endfield 1.0’s “single biggest guide misconception” (~300K), and Paik notes he is “only really efficient against bosses”
5★ worth building	XaihiAvywenna	Why — Xaihi is “the cornerstone buffer for a lot of cryo teams, if not every single cryo team”. Paik ranks Avywenna the next strongest carry after Zhuang Fangyi because she is a 5★ and her teams are F2P-friendly

Three things this table cannot tell you, and they are the reason the rest of this page exists:

1. Rankings from launch week describe a game that no longer exists. 泡椒_泡椒’s 「全角色強度及配隊推薦」 (~830K plays, 27 January) is entirely about eight operators being nerfed at public launch.
2. CN and EN disagree, and EN says CN is ahead. Kekvin’s comparison of three tier lists concludes: “CN has a head start… I value their tier list quite a lot.”
3. The letter is the wrong question anyway. Which operators have a deadline matters more — that is the next section.

You have a pile of Oroberyls, a banner with a clock on it, and one question: is this operator worth it, or do I wait? Endfield answers that more cleanly than most gacha games, because the roster splits along one hard line. Five 6★ operators sit in the pool of every Chartered Headhunting banner and always will. The other ten leave after three banners and never come back to the standing pool.

Those five are Ember, Lifeng, Ardelia, Last Rite and Pogranichnik. You can finish them at your own pace — every banner you pull is also a chance at them. Everyone else you decide about while they are still on screen, because Gryphline’s banner notices are blunt: a rate-up 6★ is “after 3 Chartered Headhunting banners, removed from the list of operators that may appear in Chartered Headhunting”, and “will also not be added to Basic Headhunting.”

So: the tier letter is the wrong question. Deadline or no deadline is the right one.

The five you can always get, and what each one is for

These five appear in every Chartered pool. None of them is a consolation prize — two of them anchor the game’s most-copied comps.

Last Rite is the boss killer, and both communities say so independently. CN’s phrase is 「對單之王」— king of single-target — from 梦轩dada’s launch-week deep dive, about 1.2 million plays. English creators arrived at the same read with a different crown: “elite single-target burst specialist” (Stix, about 70,000 views), “queen of burst damage” and “queen of cryo nuke” (GodDoggosGacha), “the Nuke Queen” (XETΛH), “deletes single targets” and “struggles in AoE” (Guobafield). Freezing is the mechanism, and that part is not a community read at all: Gryphline’s 1.1 patch notes record fixing a case where Last Rite’s combo skill “could not freeze enemies under specific circumstances.” Freeze the thing, then hit the thing while it cannot hit back.

The honest footnote is that strong and fun are different axes. Rhinne played Last Rite to max specifically to disprove his own “most boring character in the game” take, and came back with “she’s by far the most clunky, garbage, boring-ass thing in the game” — while conceding mid-video that “when you actually do damage, it is so much easier to play Last Rite.” If you bounce off her, you are not wrong and you are not alone.

Ember is a Defender who is supposed to be swinging. Her own patch-note history gives that away — the 1.1 notes fix her losing target “during her third basic attack sequence” and fix her battle skill breaking when she eats multi-hit attacks mid-cast. Defenders with a three-step combo are not standing still. She also has the game’s best redemption arc: 一树树’s「余烬风评逆转！荣登物理队的神」(about 490,000 plays, 28 January) is titled “Ember’s reputation reverses — crowned the god of the Physical team”, and 籽岷 built a stagger-focused Physical line around her with Ember as the pilot rather than the wall (about 1.17 million plays, 5 February).

What she actually feels like, from SmugSlav’s Ember video (about 20,000 views): “auto attack all day, anything the game throws at you, you press counter, you have shields for days, and you can heal… you’re just wheel chairing, man. The game is easy mode. Struggling is not even a concept if you play Ember.” Context that makes this more interesting rather than less: English tier lists rate Defenders the weakest class in the game more or less unanimously. Ember works by not really being one.

Pogranichnik holds the Physical team together — or did. The reference build is 春风满马’s “core of the Physical team” breakdown, about 1.1 million plays, and Rhinne’s EN guide is titled “He Makes Physical Teams OVERPOWERED!” (about 55,000 views, January). Two later reads walk that back, from both communities: 月下の幽冥 argued the armour-break rotation everyone copied “may be Endfield 1.0’s single biggest guide misconception” (about 300,000 plays, 10 March), and Paik’s August comp analysis says flatly that “the main problem is Pogranichnik, because he’s only really efficient against bosses — you need to invest a lot in his breach, and it only works against one enemy.” English tier lists have since moved the Physical crown to Mi Fu, who Chinese guides barely covered at all. If you are copying a January rotation in August, check it against something newer.

Ardelia is the one you already have and probably underrate. CN calls her 小羊, the little lamb; 白木哒哒哒’s clip「艾尔黛拉：让小羊来治疗你！」 is at about 2.1 million plays, and 让你爱上学习’s build guide calls her 「萌新之友」— friend of new players. She is the healing half of Nature’s two Supporters, which is why she shows up in Heat comps, Physical comps and everywhere else. Her free login-reward weapon is not universal, though: 星铁义务推广员 flagged it as close to a blank slate in a Physical team (about 370,000 plays).

Lifeng is the quiet one — a Physical Guard whose job in the standard four-Physical line is to keep debuffs on the target while someone else converts them.

One more reason to care about this group: when Gryphline ran the one-off [Fest of Brilliance] Special Headhunting in 1.2, the free 6★ selector at 120 attempts covered Laevatain, Gilberta, Ardelia and Pogranichnik — two permanents in a four-way choice. Permanents are what the game gives away.

The ten limiteds, in the order they ran

Each headlined a Chartered banner, stayed in the pool for three banners, then left. Dates are from the official notices.

Banner	Opened	Rate-up 6★	Class · Element
Scars of the Forge	Launch, Jan 2026	Laevatain	StrikerHeat
The Floaty Messenger	7 Feb 2026	Gilberta	SupporterNature
Hues of Passion	24 Feb 2026	Yvonne	StrikerCryo
River’s Daughter	11 Mar 2026	Tangtang	CasterCryo
Wolf Pearl	28 Mar 2026	Rossi	GuardPhysical
Thunder of Renewal	16 Apr 2026	Zhuang Fangyi	StrikerElectric
Fists of No Regrets	3 Jun 2026	Mi Fu	GuardPhysical
Expunger of Sin	25 Jun 2026	Camille	VanguardHeat
North Yearns the Rift Vigile	15 Jul 2026	Arcane	CasterNature
Good Morning from Your Dawnstar	9 Aug 2026	Liino	SupporterElectric

Any Chartered pool is the current rate-up plus the previous two limiteds plus the five permanents. That holds across all ten notices.

And limiteds now have a documented way back. The [Dreamscape of Wind and Snow] DEV Comm announces RE-Factor Headhunting, a new banner type opening 24 September 2026 with Yvonne rate-up. It is the first officially named rerun mechanism the game has had, which is why “gone forever” was never quite right — though it is also not a return to the standing pool, and only one operator gets a slot at a time.

What the limiteds actually feel like

Attention is a decent proxy for who people built, and Chinese single-operator guides are enormous. Here is what the biggest ones say each limited is for, with who said it.

Laevatain — the one that plays itself. 叫我棉被’s Laevatain guide is at about 2.6 million plays, the largest single-operator audience anywhere in this game’s Bilibili coverage, and 梦轩dada calls her 「輪椅一姐」— roughly, the wheelchair queen, first among comps that pilot themselves. 金乌 spends five minutes on the actual reason: she can effectively loop her ultimate (about 930,000 plays). One WeChat roundup put it as 「1托3」— she carries three passengers, so the rest of the team is whoever you own.

Zhuang Fangyi — the highest single-character claim in the game. 叫我棉被’s April guide is titled 「T0劍仙」, T0 sword immortal, about 1.4 million plays. Worth knowing it went up the week she released, when hype is loudest. 春风满马’s rotation guide for her opens with 「丟掉大腦？」— “throw your brain away?” — which tells you the play pattern is short and repeatable rather than technical.

Arcane — the flexible one. 让你爱上学习’s guide (about 1.53 million plays, 16 July) frames Arcane as 「萬能術師，一訣雙用」— an all-purpose Caster with two forms, so one pull covers two jobs. That is a real argument for a limited: not “strongest” but “fits more of your account”.

Gilberta — the plug-in. 叫我棉被 calls her 「什麼隊都能進的萬用插件」, the universal plug-in that slots into any comp (about 990,000 plays), and 让你爱上学习 sums her up as 「控拐一體」— crowd control and buff in one body. She is why Nature’s Supporters end up in everybody’s team.

Liino — the low-effort support. 让你爱上学习 filed her as 「掛機流輔助，電訣搶著要」— an idle-style Supporter that both the Electric and Arcane camps wanted (about 800,000 plays). Our Liino build page covers the three things she does not do.

Rossi — the one who broke the mono-colour habit. 叫我棉被 covered her as the enabler of 「全新混傷隊」, a new mixed-damage comp rather than another single-element line (about 590,000 plays).

Camille got 「三相之力，火隊必練」— trinity of power, mandatory for Heat — from 让你爱上学习 (about 660,000 plays). Yvonne and Tangtang are covered on the teams page, because their case is a comp, not a character.

And the free one nobody expects: the Endministrator is a genuine main carry, not a tutorial mascot. 二色彩虹’s 「管理員就是大C」 zero-spend guide is at about 880,000 plays and holds up in the standard four-Physical line.

Do not skip the 5★ shelf. Two of them turn up in top comps on both sides of the language line: Xaihi, whom GodDoggosGacha calls “the cornerstone buffer for a lot of cryo teams, if not every single cryo team”, and Avywenna, whom Paik ranks as the next strongest carry after Zhuang Fangyi precisely because “she is only a five-star and her teams can be very free-to-play friendly.” A 6★ you cannot afford to build is worth less than a 5★ you can.

Most sites have the element and class names wrong

This matters when you are cross-checking a build. Third-party databases translate Endfield’s terms loosely; Gryphline’s own operator page does not.

The rest are Electric, Nature and Physical; the rest of the classes are Caster, Guard and Vanguard. Names are the same story — these come from Gryphline’s own bilingual banner notices, not from anyone transliterating: 訣 is Arcane, 別禮 is Last Rite, 駿衛 is Pogranichnik, 萊萬汀 is Laevatain, 莊方宜 is Zhuang Fangyi, 湯湯 is Tangtang, 洛西 is Rossi, 晝雪 is Snowshine.

The roster has holes, and they decide your builds

Seventeen 6★ including the Endministrator, ten 5★, five 4★. That is small, and the distribution is lopsided enough to make your decisions for you.

Element	6★	5★ and 4★
Heat	Laevatain , Camille (Vanguard), Ember (Defender)Striker	Wulfgard , Akekuri (Vanguard)Caster
Cryo	Yvonne , Last Rite (Striker), Tangtang (Caster)Striker	Alesh , Snowshine (Defender), Xaihi (Supporter), Estella (Guard)Vanguard
Electric	Zhuang Fangyi , Liino (Supporter)Striker	Perlica , Arclight (Vanguard), Avywenna (Striker), Antal (Supporter)Caster
Nature	Arcane , Gilberta (Supporter), Ardelia (Supporter), Typhoeus (Striker)Caster	FluoriteCaster
Physical	Endministrator , Mi Fu (Guard), Rossi (Guard), Lifeng (Guard), Pogranichnik (Vanguard)Guard	Chen Qianyu , Da Pan (Striker), Purrchena (Defender), Catcher (Defender)Guard

Read the gaps and the advice writes itself:

• Heat has no Supporter and no Guard at all. A Heat team borrows its support from outside, which is exactly why the standard Laevatain comp runs a Nature Supporter in the fourth slot.
• Electric has no Defender and no Guard at any rarity, and no 6★ Vanguard. Arclight, a 5★, is the entire Vanguard answer. When Liino landed and people called her “the final piece”, CN pushed back immediately — the hole is not in the Supporter slot.
• Nature is Casters and Supporters. It supplies other teams more than it fields one.
• Physical has no Caster and no Supporter, and four of its five 6★ are Guards. Most self-contained damage core in the game, most dependent on outside help.
• Cryo is the only element with somebody in every class. If you are new and want one investment line that does not depend on winning a banner, that is the structural argument for Cryo.

What the CN scene ranked, and why it aged

None of this is our measurement. It is what large Chinese guides said, with who said it and how big the audience was, because the size of a consensus is the only thing that makes a tier claim worth repeating.

• The most-watched Endfield guide of launch week — 叫我棉被’s「這些角色必練！公測隊伍推薦榜」, about 4.4 million plays, 22 January — is a “these you must build” list organised around ready-made comps rather than a letter ladder. That is a better shape than a tier list and it is worth copying.
• 靡弦 published the closest thing CN has to a canonical launch ranking: a full-roster one-pager (about 1.52 million plays, 23 January) and an explicit 6★ strength chart (about 1.49 million, 21 January).
• Deep single-operator coverage clusters on Laevatain, Last Rite, Pogranichnik and Zhuang Fangyi. Volume of coverage is a decent proxy for who people actually built — it is not a measurement of who is strongest.
• Tier lists moved, hard. 泡椒_泡椒’s「全角色強度及配隊推薦」(about 830,000 plays, 27 January) is entirely about eight operators being nerfed at public launch. Any ranking from launch week describes a build of the game that no longer exists.

The English scene defers to CN on exactly this, and says so out loud. Kekvin’s launch-week comparison of three tier lists — two English, one Chinese — is at about 97,000 views and concludes: “CN has a head start, or a lot of the knowledge that EN may not have. I value their tier list quite a lot.”

Which makes the one concrete example of a tier list going stale worth printing. At launch, English tier lists put Ardelia two full tiers above Gilberta, and CN’s Gamer Sky list agreed — the reasoning being that one is free and one is a banner unit. By April, Guobafield had Gilberta among the best units in the game and observed she “is just getting team after team.” Both reads were sincere. Only one of them is still true. English coverage now frames the whole roster in eras rather than letters: GodDoggosGacha’s version is that “we’re safely in the second meta now — the first meta would be the heat teams, the cryo teams, with the physical teams, and now the second meta seems to point towards electric and nature.”

Where this matters for your wallet: a tier claim can cost you a limited pull, and none of the above is official. Gryphline has never published a ranking. The only thing on this page you can plan around without risk is the permanent-versus-limited split, which comes straight from the banner notices.

Two names with no banner history — and a preview that named them

Gryphline’s official operator page lists two entries that have never appeared in any headhunting pool, and lists them first, ahead of the Endministrator:

• Typhoeus — 6★ Striker, Nature, Rhodes Island; officially “huntress of the wilds”, recording the fading legends of Sami.
• Purrchena — 5★ Defender, Physical, spokesperson for the Ætherside Environmental Impact Syndrome Relief Foundation, seconded from the Talos-II General Chamber of Commerce.

The [Dreamscape of Wind and Snow] DEV Comm does not name a single operator — we read it end to end. But Gryphline ran a 1.5 preview broadcast on 21 August, and Chinese coverage of it is unanimous: 予愿安洁莉娜 posted the entrance animations for both (about 85,000 plays), 梦轩dada’s summary is titled “free max-potential new 5★ operator — Typhoeus is actually a main carry?” (about 130,000 plays), and 二色彩虹 read Typhoeus as Nature’s long-missing main damage dealer, with the caveat that she leans on her teammates to keep status applied (about 145,000 plays). Treat the specifics as reported until the version notes land with the update on 2 September; treat “these two are next” as settled. (That date was 30 August here until 24 August 2026 — see Sources.)

FAQ

Who are the best characters in Arknights: Endfield? There is no official ranking and we do not publish one. The structurally safe builds are the five permanents — Ember, Lifeng, Ardelia, Last Rite and Pogranichnik — because you can keep pulling copies of them forever. Among limiteds, the biggest Chinese coverage went to Laevatain, Last Rite, Pogranichnik and Zhuang Fangyi. That is attention, not measurement.

Which Endfield operators are permanent and which are limited? Ember, Lifeng, Ardelia, Last Rite and Pogranichnik have been in every Chartered Headhunting pool since launch. Laevatain, Gilberta, Yvonne, Tangtang, Rossi, Zhuang Fangyi, Mi Fu, Camille, Arcane and Liino are limited: per the official banner notices, each leaves the Chartered pool three banners after its rate-up and is never added to Basic Headhunting.

Do limited Endfield operators ever come back? Yes, and there is now a named system for it. The official [Dreamscape of Wind and Snow] DEV Comm introduces RE-Factor Headhunting, opening 24 September 2026 with Yvonne rate-up and a pool of Yvonne, Ember, Lifeng, Ardelia, Last Rite and Pogranichnik.

Who is the best free character in Endfield? The Endministrator. CN’s zero-spend guides build the protagonist as a real Physical carry rather than a placeholder — 二色彩虹’s「管理員就是大C」guide is at about 880,000 plays — and the standard four-Physical comp runs the Endministrator alongside Pogranichnik, Lifeng and Chen Qianyu.

What are the official English names for 庄方宜, 莱万汀, 别礼 and 骏卫? Zhuang Fangyi, Laevatain, Last Rite and Pogranichnik. Gryphline publishes banner notices in English and Chinese with the 6★ pool listed in the same order in both, which is where these mappings come from. The same source gives 訣 = Arcane, 卡繆 = Camille, 湯湯 = Tangtang, 洛西 = Rossi, 弭弗 = Mi Fu, 潔爾佩塔 = Gilberta, 伊馮 = Yvonne, 梨諾 = Liino, 晝雪 = Snowshine.

Is the element called Fire or Heat in Endfield? Heat. Gryphline’s official operator page uses Heat, Cryo, Electric, Nature and Physical, and Striker, Caster, Guard, Defender, Supporter and Vanguard for classes. “Fire”, “Ice” and “Assault” are third-party renderings.

Related

• Endfield best teams: the comps and the logic behind them
• Endfield gacha guide: banner types, pity and F2P planning
• The Liino banner and its 2 September deadline
• Liino build guide
• Endfield 1.5 roadmap
• Endfield codes and how to redeem them

Sources

Rarity, class, element and faction for every operator, and the official English and Chinese labels for each class and element, come from the data behind Gryphline’s official Operators page at endfield.gryphline.com; Typhoeus (6★ Striker, Nature) and Purrchena (5★ Defender, Physical) are listed there at positions 01 and 02 of 33. The banner table, the permanent-versus-limited rule and the [Fest of Brilliance] selector are quoted from Gryphline’s own headhunting notices on the official news hub. Kit details attributed to patch notes — Last Rite’s combo-skill freeze, Ember’s third basic-attack sequence and her battle-skill interruption — come from official version update notes, not from testing. RE-Factor Headhunting and its 24 September date come from the official [Dreamscape of Wind and Snow] Version DEV Comm, published 22 August 2026, which we read in full and which names no operator. Everything attributed to a named Bilibili creator with a play count is that creator’s read, reported as opinion, not our measurement: 叫我棉被 (launch comp list, Laevatain, Zhuang Fangyi, Gilberta, Rossi), 靡弦 (roster one-pager, launch strength chart), 梦轩dada (Last Rite, Laevatain, 1.5 preview summary), 春风满马 (Pogranichnik, Zhuang Fangyi rotation), 一树树 and 籽岷 (Ember), 白木哒哒哒, 让你爱上学习 and 星铁义务推广员 (Ardelia, Arcane, Liino, Camille, Gilberta), 金乌 (Laevatain), 泡椒_泡椒 (post-nerf ranking), 二色彩虹 (Endministrator, Typhoeus), 月下の幽冥 (Pogranichnik armour-break), 予愿安洁莉娜 (1.5 entrance animations). English-language reads carry the same status and are attributed to their creators: Stix, GodDoggosGacha, XETΛH and Guobafield (Last Rite; Gilberta; the “second meta” framing; Xaihi), Rhinne (Last Rite dissent, Pogranichnik), SmugSlav (Ember), Paik (Pogranichnik’s boss-only limitation, Avywenna), Kekvin (the CN/EN tier-list comparison and the launch-era Ardelia-over-Gilberta reading). Play counts and view counts were read from Bilibili and YouTube on 24 August 2026 and move. Coverage of Gryphline’s 21 August 1.5 preview broadcast is Chinese community reporting of a live stream, not a written official notice. Publisher and developer per the Steam listing for app 4732690 (Hypergryph / GRYPHLINE), which still reads “Coming soon” because Steam is a storefront the game has not launched on — the game has been live on Asia and Americas/Europe servers since January 2026.

Date correction, 24 August 2026. GRYPHLINE’s version overview issued to press on 21 August 2026 dates version 1.5 [Dreamscape of Wind and Snow] to 2 September 2026, reported by Gematsu and carried by Inven Global as a GRYPHLINE press release; it gives a date and no time. Checked 24 August 2026, that date does not yet appear on Hypergryph’s own news hub, whose newest bulletin is still the DEV Comm of 22 August. This page previously said 30 August, which came from a third-party schedule derivation (Game8’s) of when the version would land; the publisher’s own announcement supersedes it, and because the Liino banner is worded as running until “before version update and maintenance”, the banner runs three days longer than we said. Checked 24 August 2026.