Deadlock Record: 303,347 Players, Ahead of TF2

303,347 players online at the same time on October 16, in a game nobody can buy. This Deadlock record puts Valve’s hero shooter ahead of Team Fortress 2, and it was set by testers who got in on a friend’s invitation.

A Deadlock record that tops Team Fortress 2

The figure comes from KeenGamer (Rafly Wibowo, October 18, 2026), which credits SteamDB. The peak landed on the day Violet released. It wipes out the October 9 peak of 262,145 players, reported by Inven Global (Byungho Kim, October 10) the day Solomon arrived. Team Fortress 2 topped out at 253,997, a high dated July 13, 2023. The gap is 49,350 players.

Read that 2023 high with caution. It dates from a period when Team Fortress 2 was hit by a wave of bots (automated accounts), and none of the articles we read puts a number on their share. The Team Fortress 2 record is real, but it is not a clean yardstick. The Deadlock record has its own quirk, the opposite one: it was reached while nobody could buy the game, by invitation only.

The previous milestone, 228,086 players at the launch of City Never Sleeps, is covered in our article on the update’s six heroes. The first ranked season ends on November 5, according to KeenGamer, and Deadlock’s move to ranked will bring a different kind of data.

A vote worth 110 million ballots

Behind the Deadlock record, there is a vote. The six City Never Sleeps heroes did not come out in an order chosen by Valve. Players set it. According to the wiki’s Hero Voting page, every game played earns one vote, cast in the Hideout (the game’s lobby area). A sticker on the wall, with a pun on « I voted », is the reward.

The vote had visible effects. Rat King, the first to release on October 2, opened the ball with a 61% win rate at launch. Solomon then led the vote for what came next, as our piece on Baba and the vote explains.

What the vote reveals about players

Valve published the statistics in the October 20 announcement, titled « Shh… This Won’t Hurt a Bit. » More than 110 million votes were cast. The tightest race pitted Baba against Solomon, decided by a margin of about 0.5%.

The rest of the list says more. Only 847 players voted in every round, and only for the eventual winner. At the other end, 36,236 players voted in every round without ever picking the winner. Finally, 22,194 players voted exclusively for Nurse Harrow, and one of them cast exactly 1,000 votes for her.

Valve adds that Nurse Harrow voters have a better win rate than the voters of every other hero. The detail matters: these are players who voted for her in every game, so they are highly invested. Their success describes dedicated players, not the strength of their hero.

Nurse Harrow: four abilities and a treatment room

On October 20, Nurse Harrow closed out the City Never Sleeps release list. According to the wiki page, she is a hero who drains and heals, with a catchphrase: « Trust me, I know best. » Her infobox shows 730 health, 39.1 damage per second and 11 bullets per magazine.

• Leech Therapy: a jar of leeches that latch onto enemies, drain their health and energy, then return with the loot.
• Wretched Vapors: vapors that give allies a strong effectiveness bonus, with side effects when they wear off.
• Healing Needles: a syringe loaded with needles that inject healing straight into teammates.
• Lurking Terrors: an open bag that releases the horrors it held and forces opponents to flee.

51.0% or 48.3%: two numbers for Nurse Harrow

On October 21, a day after release, Nurse Harrow showed two win rates. Deadlock API counts 51.0% over 28,747 games, all ranks combined, ranked and unranked, across the current season. There she sits 13th out of 44, with a 0.6% pick rate. Blitz reports a 48.3% win rate, an 8.3% pick rate and a 3.4 KDA, over 1,192 games at Phantom rank and above, with a D tier.

The gap is nothing unusual. The two sites do not look at the same games. The first mixes every level, the second keeps only the best, where mistakes cost more. The Blitz sample is twenty-four times smaller. The two pick rates cannot be compared either: neither site shows its calculation method on these pages.

The per-item numbers call for the same caution. Deadlock API ranks Spiritual Overflow first with a 57.0% win rate, but over only 4,579 games, and the item is bought in just 16% of them. An item you buy when you are already winning inflates its own rate. At this stage, the 48 to 51% range describes an average hero, nothing more.