Overwatch Season 2 "Summit" first-month meta snapshot: early S-tier and why

Four weeks into Overwatch Season 2 "Summit" and the 5V5 meta has settled enough to read. This post walks through the top five heroes per role by shrunk win rate, where the dominant picks sit, and what to watch as the season moves past the early-phase data.

Season 2 "Summit" at four weeks

Season 2 launched 2026-04-14. By 2026-05-12 the community sample is big enough that day-to-day win-rate swings have mostly stopped and the tier list has a stable shape. The usual caveat applies: first-month data is noisier than mid-season data, and heroes with lower pick rates are still shrinking more aggressively toward the 50% baseline. By the halfway snapshot next week we will have a tighter read.

The rest of this post leans on shrunk win rates across 5V5, all game types, all ranks, from the current community dataset.

Tank: compressed top, no runaway pick

The Tank role is tight this season. The top five sit within a single percentage point of each other:

Reinhardt: 52.6% across ~93,000 matches. Huge sample, rock-solid placement at the top.
Hazard: 52.1% across ~39,000. The surprise of the season, level with Sigma a half point behind Reinhardt.
Sigma: 52.1% across ~95,000.
Domina: 51.8% across ~63,000.
Wrecking Ball: 51.8% across ~28,000.

No single Tank is running away with the season. That is unusual: most OW seasons have a clear top pick a full percentage point or more ahead of the pack. Summit is different: the data is telling you that at the 5V5 aggregate level, the top five Tank picks are close enough that map and comp matter more than chasing a "best Tank."

Reinhardt and Sigma both have sample sizes near 95,000 matches, which means their placements are about as confident as a tier list row can be. Hazard at 52.1% on ~39,000 is the eye-catcher: a smaller sample, but well past the noise floor, and enough to put him statistically tied with Sigma.

Damage: Torbjorn is the story

The Damage leaderboard is the opposite of Tank: it is not compressed at all. Torbjorn is running away with the role:

Torbjorn: 56.2% across ~40,000 matches. Top shrunk win rate in the entire game right now.
Symmetra: 54.0% across ~26,000.
Pharah: 52.8% across ~42,000.
Venture: 52.2% across ~27,000.
Sierra: 52.2% across ~118,000. Largest sample on the Damage list by a wide margin.

A 56.2% shrunk win rate on a hero with 40,000 tracked matches is one of the most dominant single-hero numbers we have seen since we started tracking. The shrinkage prior adds 400 imaginary 50/50 matches to every hero''s record, which barely moves the needle at this sample size: Torbjorn''s raw underlying win rate on his matches is about 56.3%, essentially identical to the shrunk number.

Symmetra at 54.0% is the second-place pick, and the fact that she and Torbjorn are both passive-damage-lean heroes suggests the early Summit meta rewards consistent chip damage against frontline tanks more than burst-ceiling carries. Pharah at 52.8% on ~42,000 matches is the consistent third option behind the two passive-damage leaders: a clear step above the chase pack and no sample-size asterisks.

Sierra at 52.2% on ~118,000 matches is worth flagging on sample size alone. She is statistically tied with Venture at the back of the top 5, but with more than four times the data behind her placement, the number is essentially locked in.

Support: Illari and Zenyatta tied at the top

The Support list has two clear leaders, then a tighter pack behind them:

Illari: 53.8% across ~95,000 matches. #1 by a tenth of a point.
Zenyatta: 53.7% across ~100,000. Statistically tied with Illari on an even bigger sample.
Brigitte: 52.1% across ~43,000. The quiet riser of the season.
Mizuki: 51.5% across ~132,000 tracked matches. The highest sample on the role outside the top two, meaning his placement is essentially locked in.
Lucio: 51.5% across ~60,000.

Illari and Zenyatta are functionally tied. With 95K+ matches behind both numbers, the 0.1pp gap is noise: either is a defensible "main this" pick at the top of the role. Brigitte at 52.1% on ~43,000 is the surprise #3, sitting a clear step behind the leaders and a clear step ahead of the Mizuki/Lucio cluster. Mizuki''s huge sample and 51.5% placement means he is the safest "I just want a reliable Support" pick this season: high pickrate, just above baseline, low variance.

Heroes that moved

A few season-to-season shifts worth flagging. The sparsity caveat applies: Counterwatch''s tracked DAU has grown roughly 5× over the past year, so cross-season pickrate deltas read as directional rather than precise. Shrunk win-rate comparisons within Summit are fine; comparing to Vendetta-era (Season 20) numbers needs the caveat.

Torbjorn''s climb is the headline. His 56.2% is not a number that has been available to Damage players in prior recent seasons. That could be a pure Summit-patch effect, or a community-learning effect with players getting better at Torbjorn setup and turret placement in the current map pool. The halfway snapshot will tell us which.

Hazard at 52.1% on ~39,000 matches is the other Tank story. He has historically been a niche pick, and seeing him level with Sigma at the top of the shrunk leaderboard, on a sample that is now well past the noise floor, is the most interesting Tank shift since the start of the year. Whether that is sustainable through a balance patch or a function of the current map rotation is the thing to watch.

Wrecking Ball''s presence in the top 5 Tanks at 51.8% with 28,000 matches is also notable. Ball typically lands mid-tier in 5V5 because the single-tank economy punishes his "dive and disappear" playstyle. Summit''s early data suggests either a kit adjustment or a meta reshaping is letting him reach his ceiling more consistently.

Confidence caveat

Four weeks in is still early. Mid-season balance adjustments can flip the list overnight: a single Torbjorn nerf would collapse the Damage leaderboard, and a Reinhardt or Hazard adjustment would reorder the Tank top 5. We will publish a halfway snapshot on 2026-05-19 to see how much has shifted.

The sample-size story also matters. Torbjorn at 56.2% on 40,000 matches is high-confidence. Sierra at 52.2% on ~118,000 and Pharah at 52.8% on ~42,000 are also high-confidence placements. Lower down the list, sample sizes drop and shrinkage does more work: placements below #5 per role can move by a full percentage point between snapshots without necessarily signaling a real meta shift.

What to watch in the halfway snapshot

Three specific things the next two weeks of data should answer:

Does Torbjorn''s 56% edge survive a patch? His lead is wide enough that it is unlikely to flip without a direct nerf. If the edge holds for another four weeks, it is a "learn this hero" moment for Damage players.
Do the compressed Tanks sort out? Tank roles usually settle into a clearer #1 by mid-season. If Reinhardt''s lead stays this thin and Hazard holds level with Sigma, it is a sign the patch balanced Tanks unusually well.
Where does Mizuki land on his growing sample? At ~132,000 matches and climbing, his placement is essentially immovable. But watch whether his shrunk number drifts up or down: another half percentage point and he is in the Brigitte conversation rather than the Lucio one.

For the live data at your specific rank, filter the full tier list to your division. The "All" view smooths across skill levels and often buries role-specific rank metas. For the role-by-role "best pick to main" rankings with consistency and sample size factored in, the best one tricks page is the answer. For the full write-up on shrinkage, per-rank filters, and refresh cadence, see the methodology page.

The Counterwatch app runs these numbers live in your match. When the enemy team locks picks, the overlay scores your current lineup against the live enemy comp using the same tier data you just read above.