Valorant Agent Tier List, Patch 13.05 (September 2026)

Riot does not publish a tier list, pick rates or win rates for Valorant, so this one is ours and the reasoning is written out line by line. It is built for patch 13.05, live since 1 September 2026, and it ranks agents inside their role rather than against the whole roster, because your team fills a controller slot and an initiator slot long before it argues about the best agent in the game. The short version of the last three months: Riot touched Iso, Yoru and Phoenix and nobody else, so the roster has barely moved.

Start with the row that decides your composition — the controller.

Short answer

Ranked for solo queue on patch 13.05, on the pool that changed in 13.04 when Abyss entered the competitive queue and Breeze left.
Controllers first: Omen and Clove are the two we would lock without knowing the map.
The only balance changes since June hit Iso (instant re-equip), Yoru (Gatecrash beacon 15 → 20 seconds) and Phoenix (ultimate cost 6 → 7 points).
Miks and Veto are deliberately unranked — we have no data on them and will not invent it.
Disagree in a picture: load the Valorant roster in our tier list maker, drag it your way and send the link.

How this list was built

Three lines, because a tier list without them is unreadable.

What is being ranked: how much an agent contributes that the team cannot get from anyone else. Not how fun they are, not how high their ceiling is in a five-stack with voice comms.

Who it is for: ranked solo queue, average to high ladder, one agent per role. If you play in a coordinated five, Astra and Harbor climb and Reyna falls, and that is not a flaw in either list — it is a different question.

When it is true: patch 13.05, 1 September 2026. Riot's own notes promise a heavier patch in 13.06 with Agent Gear changes, so treat the date as a shelf life, not decoration.

Everything below is our reading of ability design plus the published patch history. Where we have neither, we say so instead of filling the gap.

Controllers: the row that decides the composition

Agent	Row	Why
Omen	S	Smokes he can place from anywhere on the map plus two ways to reposition; the pick that never wastes a slot.
Clove	S	The only controller who can still smoke after dying, which in solo queue is where half your smokes are needed.
Brimstone	A	Cleanest smokes for a site take and a real ultimate, but no mobility — punished on maps with long rotations.
Viper	A	On the biggest maps her wall is close to a sixth player; on small ones she is a worse Brimstone.
Astra	B	Highest ceiling on the list and the one that collapses fastest without callouts.
Harbor	B	Water walls block vision and bullets, which is strong on paper and awkward in the middle of a rushed default.

One controller per team is the floor, not the target, and the role goes unfilled more often than any other. To stop losing rounds to nobody smoking, learn Omen first and Brimstone second: between them they cover the whole map pool, and neither asks anything of your teammates.

Astra sits in B for one reason: her value is entirely in setups the team executes, and a stranger's default is not a setup. In a five-stack she is a different agent, and we would put her in S.

Duelists, initiators and sentinels

Duelists. The role that gets picked instantly and understood last.

Agent	Row	Why
Raze	S	The most damage utility in the game; wins site takes and post-plants without needing to hit a shot.
Jett	S	The only duelist who can take an Operator angle and leave the fight she started.
Neon	A	Fastest entry on the roster; the speed is only worth it if someone is behind you.
Reyna	A	Heals and escapes off her own kills and contributes nothing to anyone else — a solo-queue engine, a poor team pick.
Phoenix	A	Self-sufficient flashes and heal; his ultimate now costs 7 points instead of 6 after patch 13.02.
Iso	B	Faster since 13.01, when equipping a weapon after Double Tap became instant, but still narrow.
Yoru	B	Buffed in 13.01 and still a specialist: the reward is high, the failure mode is dying alone behind the enemy team.
Waylay	B	A 2026-era duelist we have not seen dominate a patch; ranked low on evidence we do not have, not on evidence against her.

Initiators. The role that quietly wins rounds.

Sova stays at the top because Recon Bolt gives information a team can act on immediately, and KAY/O joins him because suppression is the only tool in the game that turns abilities off. Fade, Breach and Skye all sit in A — Fade for information plus a Prowler that opens a site, Breach for stacking flashes and stuns through walls on the maps with tight corridors, Skye because a flash, a heal and a dog is a full toolkit for a player who wants to do everything.

Gekko is B for one honest reason: his utility comes back, which forgives mistakes, and forgiveness is worth less at the top of the ladder than a Recon Bolt that lands. Tejo we would put in A with a caveat — his kit throws utility from across the map, which is exactly what teams want and exactly what is hardest to judge from the outside.

Sentinels. Cypher and Killjoy are S. Cypher gives you the enemy's position twice — once from a trap, once from his ultimate — and Killjoy locks a site so hard that two players can hold it. Vyse and Sage are A: Vyse denies an entry, Sage denies a push and brings someone back. Deadlock and Chamber are B. Chamber is not a bad agent; he is a rifle with a teleport, and if your aim is not the reason you win rounds he gives you less than any other sentinel.

What did patch 13.05 actually change?

Almost nothing, and that is the useful part.

Riot shipped 13.05 on 1 September 2026 and described it in their own notes as a small patch. No agent balance changes. What did ship: audio adjustments to sound sprays, Pick'Ems for Champions Shanghai, penalties for stream sniping and for instalock tools, and bug fixes for Chamber, Cypher, Deadlock, Fade, KAY/O, Omen, Sage and Waylay. The notes also flag that Agent Gear changes are coming in 13.06.

The changes that do matter are older:

Patch	Date	What changed
13.01	14 July 2026	Iso equips his weapon instantly after Double Tap; Yoru's Gatecrash beacon lasts 20 seconds instead of 15, and Fakeout clones hold his last equipped weapon rather than his strongest.
13.02	28 July 2026	Phoenix's ultimate costs 7 points instead of 6. Split, Corrode and Lotus added to Retakes.
13.04	18 August 2026	Abyss entered the competitive queue, Breeze left it. No agent balance changes.
13.05	1 September 2026	Bug fixes and player-behaviour penalties. No agent balance changes.

Three agents adjusted in three months is a quiet stretch by Valorant's standards. If a tier list claims the meta was rebuilt this Act, ask which patch note says so.

That same standard applies to us, which is why two agents are missing from every table above.

The agents we did not rank. Miks (Controller — M-pulse, Waveform, Harmonize, Bassquake) and Veto (Sentinel — Interceptor, Crosscut, Evolution, Chokehold) are both on the playable roster, and both are absent from every row above.

We have their ability names from the public agent API and nothing else: no balance pass in patches 13.00 through 13.05, no published pick data, no hours of our own on either. Putting them in B because that feels safe would be exactly the move this site keeps calling out in other people's lists.

If you main one of them, your placement is better than ours. Build the board in the tier list maker — the Valorant preset loads all 29 agents, spelled as the game spells them — and the link carries your layout, so you can send an argument instead of a paragraph.

Do you actually need a tier list in solo queue?

Not in the way most people use one. In solo queue the agent select screen is decided by whoever clicks first, and a tier list does not survive contact with three instalocked duelists.

Where it does help is the opposite direction: knowing which two agents you would be happy to fill with. A player who can smoke competently and hold a site competently never has a bad agent select, because the roles nobody takes are exactly those two. That is the practical use of ranking the roster — finding your second and third pick, not your first.

And if you are picking for fun rather than for rank, the agent randomiser settles it in one click, and our agent quiz at least gives you someone to blame.

What usually goes wrong

A tier list is copied straight into ranked. Symptom: you lock the S-row duelist, go 4-16 and conclude the list is wrong. Cause: a row is a statement about an agent's ceiling with a team behind it, not about your next match. What to do: pick the highest-ranked agent in the role your team is missing, which is almost never duelist.

The patch is not written down. Symptom: an argument about Phoenix that neither side can resolve. Cause: one player is describing him before patch 13.02 and the other after, and the ultimate cost moved from 6 to 7 in between. What to do: put the patch number in the image, the way the tables above do.

Win rates get quoted with no source. Symptom: a percentage in a comment that nobody can trace. Cause: Riot publishes none, so every number in circulation comes from a third-party sample with its own rank distribution and time window. What to do: ask which site, which rank band, which patch — and if all three answers are missing, ignore the number. Our guide to building a tier list that survives a reply goes through the same trap with real threads.

Astra and Harbor get judged by solo-queue results. Symptom: a coordinated team's best controller looks terrible on the ladder. Cause: both need a call before the utility is worth anything. What to do: rank them twice, once for stacks and once for solo, and say which list you are showing.

The map pool is out of date. Symptom: an agent recommendation built on Breeze, which left the competitive queue in patch 13.04. Cause: map-specialist picks age faster than anything else in the game. What to do: check the current pool before you build the list, and note the rotation date next to it.

Take it apart

Your controller row is the one worth arguing about. Drag the 29 agents into your own board in the tier list maker, keep the patch number visible, and post the link.

If you want your crosshair and mouse to match the agent you land on, our sensitivity guide, the crosshair codes page and the sensitivity converter are the next three stops. Ranking guns instead of people happens in the CS2 weapon tier list; the battle-royale version is the Apex legend tier list.

FAQ

Is this the official Valorant tier list?

No. Riot does not publish a tier list and does not publish public pick rates or win rates, so every agent tier list you have ever read is somebody's opinion with a roster attached. This one is ours, it is built for patch 13.05, and the reasoning behind each row is written out so you can disagree with a specific line instead of the whole page.

Which agent should I pick if I only ever play solo queue?

A controller you can smoke with confidently, or Reyna. Those are the two picks that do not depend on a teammate cooperating: a controller is the role that goes unfilled most often, and Reyna heals and escapes on her own kills without needing anyone to trade or follow up. Both work with zero communication, which is the actual constraint in solo queue.

Did patch 13.05 change the meta?

Barely. Riot called it a small patch and shipped no agent balance changes at all, only bug fixes for Chamber, Cypher, Deadlock, Fade, KAY/O, Omen, Sage and Waylay, plus penalties for stream sniping and instalock tools. The notes also say a heavier patch is coming in 13.06 with the Agent Gear changes, so expect this list to move then.

Why are Miks and Veto not in the tiers?

Because we cannot rank them honestly yet. Both are 2026 additions, we have no measurements of our own and Riot has published no balance pass on either, so any placement would be a guess dressed as analysis. They are listed separately with their ability names and nothing more.

Does the map pool change which agents are good?

Yes, and it changed recently: patch 13.04 put Abyss into the competitive queue and pulled Breeze out. Breeze rewarded long-range control and a wall-heavy controller; agents whose value comes from covering huge open sightlines lose one of their best stages when it rotates out. Check the pool before you commit to a one-map specialist.