FRITZ 20 launch: the chess engine shifts from competition to personalised training

Bonn, 5 December 2026. Vladimir Kramnik, the reigning world champion, sat motionless at the board after game six. Deep Fritz won the match 4-2, and for the first time in chess history a world champion lost to a machine under classical time controls. I replayed all six games in a small cafe in Hanoi, writing every move into a squared notebook. What haunted me was not the human defeat. It was the machine's silence: it did not celebrate, it simply calculated.

Twenty years later, the software line called Fritz returns with a very different claim. No longer a challenger. FRITZ 20 presents itself as a training revolution — your personal chess trainer, your toughest opponent, your strongest ally. It targets both players taking their first steps into serious training and players already competing at tournament level. On the surface it is the familiar marketing language of every sports product. But it lands on the sorest spot of semi-serious chess: many people own tools, very few own a method.

Fritz is not a new name. The first version appeared in 2026, developed by Frans Morsch and Mathias Feist, and was brought into the ecosystem of ChessBase, the chess software company based in Hamburg, Germany. For two decades Fritz was the public face of computer chess: in 2026 Deep Fritz drew 4-4 with Kramnik in Bahrain; in 2026 X3D Fritz drew 3.5-3.5 with Garry Kasparov. Then came the era of Stockfish, Leela Chess Zero and NNUE neural networks, when engine strength soared to the point where human-versus-machine matches stopped being interesting as sport.

The numbers make it clear. Top engines now exceed 3600 Elo on computer rating lists, while the highest human rating remains Magnus Carlsen's 2882, set in May 2026. That gap of nearly 700 points will not close. So the question of this decade has changed: the machine no longer needs to prove it is stronger than us, it needs to prove it can teach us.

That is exactly where FRITZ 20 places its bet. The value of a training engine lies not in its calculating power, but in its ability to translate the machine's language into human language. I call this the explanation gap. A strong engine tells you that this move leads to a 1.4 advantage; it does not tell you why, does not tell you where you went wrong three moves earlier, and does not tell you what to train over the next two weeks. Based on my experience covering matches during four years of chess commentary for VTC, I learned that viewers do not need to know what the engine thinks, they need to know what the player thinks. A decent training programme must do the reverse of commentary work: explain the machine to the human.

Error diagnosis, first of all, must work by pattern rather than by single game. A 1600-level player makes the same kind of mistake hundreds of times: missing an intermediate move, or trading pieces too early before finishing development. A familiar example is trading on f6 to damage the pawn structure; the engine shows the evaluation drops 0.3, but it does not say the real problem is that you never completed your queenside development. Good software must group those errors into a repeatable technical profile instead of letting them scatter like the accidents of one evening.

Difficulty must also be adjustable. You need to win often enough to stay motivated, but not so easily that you stop learning. This is where pure engines fail, because they do not know how to hold back at the right moment.

An opening repertoire tied to style is a matter for the learner, not for the database. A player who loves structure learns different systems from one who loves attack. If the software only offers the highest-rated lines, it turns you into a copy of a data file, and turns your games into a translation with no voice.

Then there is opponent simulation. Most amateur players do not lose because they face a 3600 engine, but because they face a 1750 opponent who specialises in trading down into an endgame. Training against the exact type of opponent you meet at local tournaments has more practical value than training against a perfect machine.

A spaced repetition exercise system matters just as much. Analysing a game without returning to it two weeks later means most of the lesson evaporates, and that is why many players read thousands of games without improving.

Data from online chess platforms reveals a familiar paradox: players below 1800 Elo spend most of their self-study time on openings, while their errors cluster in the middlegame and endgame. That is why a strong engine does not automatically produce a strong player. A tool only amplifies the method you already have; if the method is wrong, it amplifies the waste as well.

Software costs money. It needs a computer with enough power, a stable connection, and above all spare time. In many places, Vietnam included, the biggest gap for female players is not software but coaching networks and invitations to tournaments. In the summer of 2026 in Russia, I went looking for queens whose names were not on the media map, while the whole country watched only the men's football World Cup. The lesson from that project still holds on a 64-square board: a tool only matters when it comes with a chance to compete.

The risk of stylistic homogenisation is no less serious. When everyone learns from the same evaluation source, junior games become so similar they turn dull, and options dismissed as inaccurate are discarded before anyone tests them. Judit Polgar beat Kasparov in Moscow in 2026 and was once ranked eighth in the world with 2735 Elo. She did not become an elite player by optimising every move, but by daring to play moves the machine considered imperfect yet opponents did not want to face.

And there is one thing no software can solve: courage. On the chessboard the only boundary is the boundary of skill — an engine does not distinguish by gender, nationality or skin colour. But the training market does. Hou Yifan once reached 2686 Elo and competed in open tournaments against male grandmasters; the notable thing is not which software she used, but how many tournaments she was given to prove herself.

The value of a player is not written in the price tag of software, but in the fate of the girls who dare to sit down at the board. Picture an afternoon when your younger sister sits at a chessboard without asking anyone's permission, with a patient teacher inside her computer who never runs out of time. I have seen a generation of young players ready to learn through data, and I believe they will not step back. What is still missing is not a stronger engine, but an answer to a very old question: who gets to sit in that chair?