A somewhat calmer day, with three draws.
Keymer chose the Karpov variation in the Nimzo (9…b6). With 12…exd5, he avoided the sharp 12…Bxc3 13.dxe6 Bxf3 14.gxf3 fxe6, where it seems that black equalises with exact play. Instead, Gelfand got a slight initiative, but Keymer defended well, and the game faded into a drawn endgame.
As white in the Ruy Lopez, Svidler sacrificed a pawn, and after 18.Qc3, it seemed like black was in trouble.
After thinking for almost 40 minutes, van Foreest found the best defence in 18…b4 19.Qf3 Rb8. White got his pawn back, but black managed to castle, and after that, the game was headed for a draw.
Grandelius once again chose the Slav Defenece, and seemed to be ok, but after the risky 17…e5 (instead of 17…Bd6), Mishra had the initiative. Maybe he should have tried 23.cxd4 instead of taking with the bishop. 23.cxd4 allowed Grandelius to play some exact moves and reach an easily drawn endgame with opposite-colored bishops.
After a calm start, the game got very complicated. It seemed that Gukesh had planned to answer 32.Qxh6 with Rd6, but discovered that this would run into 33.Bxf7+.
This meant that he had to re-calculate, putting him in severe time trouble. This made it virtually impossible to find 33…Nxh3 34.gxh3 Rh4, a continuation that would have allowed him to play on. Instead, Erigaisi found an exchange sacrifice that gave him an unstoppable mating attack.