Hikaru's Blitz Rating Record Was 3470, Then 3472, Now 3490: What Chess.com's Own Data Shows

What is Hikaru’s blitz rating record right now?

Hikaru’s blitz rating record is 3490, reached at 20:13 UTC on 3 October 2026, according to the best-rating field in Chess.com’s public API. His live rating when we last read it, at 21:53 UTC the same evening, was 3490 again, after a draw and seven more wins.

Those numbers come from the stats endpoint for his account (api.chess.com/pub/player/hikaru/stats). Its blitz best block held 3490 with the Unix timestamp 1791058418, which is 20:13:38 UTC on 3 October 2026.

One quirk: the game link inside best opens Nakamura’s win over Magnus Carlsen on 24 September, not the record game. Chess.com’s API documentation calls that field the “URL of the best-win game”, which we read as his best win by opponent.

To find the record game we followed his rating through the monthly game archives on the same API. The peak came from a 34-move win with White over FM Edgar Andreasyan, who plays as BerlinNoMercy.

Is that the highest blitz rating on Chess.com by anyone? Chess.com called 3470 “a Chess.com record” on 29 September, and Nakamura has only gone higher since. Among the 50 players on the API’s blitz leaderboard that evening, the next best peak was Carlsen’s 3412. We found no current official all-time list to check it against.

3470 or 3472: why did Chess.com and GothamChess give different numbers?

Chess.com and GothamChess gave different numbers because each described the record at a different moment on the same day, 29 September 2026. The game archive shows 3470 at 03:31 UTC and 3472 at 15:27 UTC.

Chess.com’s recap of Titled Tuesday, written by Nathaniel Green, says Nakamura “reached a Chess.com record 3470 blitz rating just before the event started”. The archive agrees: 3470 came from a three-minute win about eleven and a half hours before his first tournament game finished.

In the tournament itself, his first win left the rating at 3470, the second made it 3471, and the third, against GM Valery Kazakouski, made it 3472 at 15:27 UTC. That is the figure in GothamChess’s video, where Levy Rozman says Nakamura is the one person who “has reached 3,472 on chess.com”. The game’s own page agrees: “GM ContrVersia (3055) vs GM Hikaru (3472)”.

Three draws and a loss then took him from 3472 down to 3442 in about an hour, and the recap mentions only the earlier figure.

The 3472 mark was matched again at 05:17 UTC on 3 October and beaten at 14:25 UTC. It had stood for three days, 22 hours and 58 minutes, so it fell in just under four days. The video went out at 17:00 UTC on 2 October, and its number stopped being the record less than 22 hours later.

The games behind the Hikaru blitz rating record

On 15 September, after three losses in that day’s Titled Tuesday, the archive has Nakamura at 3342. Eighteen days and 253 rated blitz games later he was at 3490, with 231 wins, 12 draws and 10 losses.

Rozman’s video walks through four games. Here is what the archive says about each, plus the day the video missed.

1. Tan Zhongyi, 17 September: a win worth nothing

Chinese grandmaster Tan Zhongyi was Nakamura’s first opponent in a 3 0 Thursday tournament on 17 September. She resigned after 33 moves of a Sicilian Kan. Rozman says he thinks this game took Nakamura past 3400. The archive shows something odder: he had passed 3400 the day before, and his rating was 3410 before the game and 3410 after it. Beating a grandmaster rated 2779 earned him nothing.

2. Alexander Rustemov, 23 September: fourteen wins, one loss

Rozman guesses Nakamura “went 14-1 or something” against GM Alexander Rustemov, and for 23 September the archive shows exactly that: fifteen three-minute games, fourteen wins, one loss. Each win paid two points and the loss cost fourteen, so a 14-1 day against a grandmaster rated around 3050 was worth a net fourteen points and left him on 3443.

3. Magnus Carlsen, 24 September: the win worth seven

The win over Carlsen came in a 3 0 Thursday tournament on 24 September. Carlsen chose a Sicilian with the offbeat 3…f6, Nakamura hit back with 12.d5 and won the exchange, and Carlsen resigned on move 58 with both players under ten seconds. Nakamura went from 3433 to 3440 and Carlsen finished on 3394. Seven points was the largest gain from any win in the run, because Carlsen was the only opponent rated close to him.

4. Valery Kazakouski, 29 September: the 3472 game

The game that made 3472 was Nakamura’s third of Titled Tuesday on 29 September: five minutes each, Black against GM Valery Kazakouski, who plays as ContrVersia under Lithuania’s flag and was rated 3055. It was a King’s Indian. Rozman calls the position after 21…d5 equal, and Kazakouski resigned eight moves later, after 29…Ndxf2. The win was worth a single point, 3471 to 3472.

5. 3 October: 34 wins, 34 points

The video was recorded on 2 October, a day too early for this. On 3 October Nakamura played 34 rated three-minute games between 01:01 and 20:13 UTC, against seven titled opponents rated between 2880 and 2913, and won every one. Sixteen were against the same FIDE master. Each win moved the rating by exactly one point, from 3456 to 3490. With the ten before them, that made 44 straight wins.

Hikaru’s blitz rating history: the key records in one table

Earlier records come from Chess.com’s own reports, recent ones from the game archive. All times are UTC.

The archive shows smaller new highs between the rows, such as 3431 in August 2025 and 3465 in May 2026. The July 2025 report puts it drily: “The previous record-holder was… Nakamura at 3408.”

Carlsen held the record for a while. According to the 2024 report, he took it in December 2023 and pushed it into the 3370s in January 2024. Nakamura’s archive has him back above that on 4 February, five days before he broke 3400. From 3336 to 3490 is 154 points in under three years, and the last 21 arrived in five days.

How close is 3500, really?

On paper 3500 is close: ten points above the 3490 peak, which was also Nakamura’s live rating when we read the API at 21:53 UTC on 3 October 2026. At his level that means a long run of wins with no slip.

Here is what results were worth across his 281 rated blitz games from 14 September to the peak, taken from the archive:

• Opponent rated under 2800: a win paid nothing (7 wins).
• Opponent in the 2800s or 2900s: a win paid one point in 126 of 146 cases, a draw cost seven and a loss cost 15.
• Opponent in the 3000s: a win paid two points in 79 of 83 cases, a draw cost six and a loss typically cost 14.
• Opponent rated 3100 to 3299: a win paid two to five points, and a loss cost 12 to 14.
• Carlsen: seven points.

So against the 2900-rated players he met on 3 October, one loss wipes out fifteen wins and one draw wipes out seven. The draw at 21:14 UTC, against a candidate master rated 2916, did exactly that.

None of this is new. In 2023 Chess.com described him “arduously picking up a single rating point for each victory” (the archive has ten of those fifteen wins at two points each), and its 2024 report warned that “losses are costly”.

From 3490 there are three realistic routes to a Hikaru 3500 rating:

1. Ten one-point wins in a row against 2900-rated opposition.
2. Five wins in a row against players rated in the 3000s, if those still pay two points each at this height.
3. A strong tournament, where wins pay more but the opponents can bite, as 29 September showed.

The 3500 in GothamChess’s title is a prediction, and should be read as one. In the video Rozman says: “I cannot wait for us to have the first 3500 blitz rated player on chess.com.” The archive supports either outcome: 34 points gained in a day, and 30 lost in an hour.

Is a Chess.com blitz rating the same as a FIDE Elo?

No, a Chess.com blitz rating is not the same as a FIDE Elo. Nakamura’s official over-the-board blitz rating on his FIDE profile was 2800 when we checked on 3 October 2026, which is 690 points below his Chess.com peak.

The video’s title says “3500 Elo”, but strictly the number is not an Elo rating. Chess.com’s help centre says: “Chess.com uses the Glicko rating system”, which adds a rating deviation (RD), a measure of how sure the system is about your rating. Nakamura’s RD in the API was 32. In Mark Glickman’s paper on the system an unrated player starts with an RD of 350, and “A low RD indicates that a player competes frequently.”

Chess.com’s own explainer calls Glicko “a modified version of the Elo system”. Glickman, who created it in 1995, puts it the other way round: “The Elo system, coincidentally, turns out to be a special case of my system.”

The formula matters less than the pool. A rating only measures you against the other people in the same pool, and FIDE blitz and Chess.com blitz are different pools.

Carlsen is the clearest example. His FIDE blitz rating was 2860 on 3 October 2026, which is 60 points above Nakamura’s. On Chess.com his peak is 78 points below Nakamura’s. Same two players, opposite order.

For scale at the other end of the ladder, the fruit fly brain map that was turned into a chess player reached a reported 700 Elo.

Hikaru vs Magnus: the gap on 3 October

On Chess.com’s API on 3 October 2026, Carlsen’s best blitz rating was 3412, set on 30 September and 78 points short of Nakamura’s, and his live rating was 3400.

Head to head is a different story. In the archive, the two main accounts met in 41 rated blitz games between October 2023 and 3 October 2026. Carlsen won 13, Nakamura won 10 and 18 were drawn.

So “Hikaru beats Magnus” is true of 24 September and of the rating list, but not of their rated blitz games since late 2023. The difference on the list has to come from results against everyone else, and Nakamura plays far more of those games: more than 45,000 blitz games on his account’s record, against fewer than 7,000 on Carlsen’s.

Rozman’s view that Carlsen is “not as much of a grinder as Hikaru” is an opinion, though the game counts point the same way.

What the GothamChess video gets right, and what is a prediction

The video is “Hikaru Just Solved Chess (3500 Elo)”, published on 2 October 2026 on a channel with about 8.2 million subscribers. It had 669,884 views when we checked at 21:45 UTC on 3 October. Checked against the archive, it mostly holds up.

• Fact, confirmed. 3472 was reached against GM Valery Kazakouski, rated 3055, and the Carlsen game ended with the ratings Rozman quotes.
• Not supported. The Tan Zhongyi game did not take him past 3400, which Rozman hedged with “I think”. He had crossed it a day earlier.
• Title, not content. Nobody solved chess, and nobody had reached 3500 in blitz when we checked. The number in the video is 3472, and the scale is Glicko.
• Prediction. The first 3500 blitz rating is something Rozman expects, not something that has happened.

The same channel posted a Short the next day titled “Vote Solomon Deadlock”, about a chess-themed hero Valve has announced for Deadlock. We cover that one in our piece on Deadlock’s Solomon.

What a chess rating teaches about a stock figure

Nobody types Nakamura’s rating in. Every game is logged in one place with a timestamp, and the rating is worked out from that log each time a game ends. Chess.com’s recap, GothamChess’s video and this article are all copies of the number, and each copy started going stale the moment it was made.

Plenty of businesses run on the copies. A stock figure sits in a spreadsheet because somebody typed it on Monday. By Thursday the warehouse, the sales desk and the accountant each hold a different version, and every one of them was correct at some point.

Arguing over which is right is the 3470 versus 3472 argument, and it has the same answer: go to the log, not to the last snapshot.

In the systems we build, a stock figure is calculated from logged receipts, picks, adjustments and despatches, so the number carries its own history and nobody has to remember to update it. That is what a proper stock record is, and you can click through working examples in our demos.

OpsMavix builds custom internal systems for inventory, orders, purchasing, production and reporting, replacing spreadsheets and disconnected tools with one place where the number lives.

FAQ

What is Hikaru’s highest blitz rating on Chess.com?

Hikaru Nakamura’s highest blitz rating on Chess.com is 3490, reached at 20:13 UTC on 3 October 2026, according to the site’s public API as read at 21:53 UTC that day. He had been playing minutes earlier, so check his stats page for the current figure.

Has anyone reached 3500 on Chess.com?

Not in blitz, as of 21:53 UTC on 3 October 2026. The highest blitz rating we could find is Nakamura’s 3490. Bullet is a separate rating, and there Nakamura’s own best is 3570, set on 11 November 2020, according to the same API.

Did Hikaru beat Magnus Carlsen during the record run?

Yes, once. Nakamura beat Carlsen with White in 58 moves in a 3 0 Thursday tournament on 24 September 2026 and gained seven points. It was their only rated blitz game on the main accounts between 15 September and 3 October.

Who did Hikaru beat to reach 3472?

GM Valery Kazakouski, who plays on Chess.com as ContrVersia. It was Nakamura’s third game of Titled Tuesday on 29 September 2026. He had Black in a King’s Indian, Kazakouski resigned after 29 moves, and the rating went from 3471 to 3472.

Why does Hikaru only gain one point for a win?

Because he is rated hundreds of points above nearly everyone he plays, so a win is the expected result. Chess.com’s help centre says: “As players improve and reach higher levels on Chess.com, the rating changes after each game become smaller.”

Final takeaway

The honest answer to “what is Hikaru’s blitz rating record” is a number with a time attached: 3490 at 20:13 UTC on 3 October 2026, and possibly more by the time you read this. 3470 and 3472 were true as well, for about twelve hours and about four days. For today’s figure, read the source.

The same rule holds for any number a business runs on. If your stock, order or production figures depend on who you ask and when they last updated a sheet, the fix is one log that works the number out for you. Book a free Operations Leak Audit and we will show you where your figures drift apart.