Chess bots to play against, from beginner to grandmaster
Twenty rivals with a name, a rating and a way of playing, from one who hangs pieces to one who does not forgive anything. Pick the one just above you, and after every game the app shows you what went wrong.
The twenty rivals
Five levels, four rivals each. The rating is the strength each one plays at; beat one a few times and move up.
| Level | Rival | Rating | How they play |
|---|---|---|---|
| Beginner | Beto | 750 | Just learning. Hangs pieces. Don't feel bad if you win in 10 moves. |
| Lola | 900 | Still learning the openings. Leaves pieces where they can be taken. | |
| Ana | 1100 | Slow and careful. Avoids trades unless they help. Hard to rush. | |
| Tomás | 1150 | Likes good knight squares and central pawns. Will outplay you slowly if allowed. | |
| Intermediate | Carmen | 1250 | Club beginner. Spots one-move threats and misses the rest. |
| Diego | 1350 | No big style: plays the position as it comes. A real tester at this level. | |
| Sara | 1450 | Quiet manoeuvres, slow improvements. Don't expect tactical fireworks. | |
| Hugo | 1550 | Solid club level. Rarely hangs a piece, but two-move tactics get past him. | |
| Advanced | Kenji | 1670 | Defends calmly and counterattacks when you drop your guard. |
| Tariq | 1760 | Trained on real club games. Solid openings, and the odd tactic slips past. | |
| Ingrid | 1870 | Clean structures and few mistakes; you have to beat her move by move. | |
| Pavel | 1950 | A complete player; punishes the slightest inaccuracy. | |
| Master | Iván | 1750 | Strong club player. Punishes clear mistakes and drifts in quiet positions. |
| Marta | 1850 | Builds her position brick by brick. Punishes inaccuracies in the long run. | |
| Javier | 1950 | Just below expert level. You need a real plan to beat him. | |
| Gabriel | 2150 | All-round strong player at the gates of master level. No weak spots. | |
| Grandmaster | Levon | 2600 | Grandmaster strength. One clear mistake is usually enough to lose. |
| Judit | 2650 | Rarely leaves you anything. | |
| Hikaru | 2792 | Tactics and speed. Punishes the smallest inaccuracy. | |
| Magnus | 2841 | Endless technique. The last boss. |
Bots that lose like people, not like broken engines
The usual way to make an engine weaker is to make it play a random bad move now and then, which feels nothing like a real opponent. Chess Coach does it differently: the mistakes a rival makes are chosen among the moves the engine itself was considering, so they are the plausible slips a player of that rating makes.
The four advanced rivals are Maia, a neural network trained on millions of real online games to predict the move a player of a given rating would make. They open, defend and blunder the way club players do. The master and grandmaster rivals are full engines, limited only as much as their rating needs.
After the game
Every game against a bot goes to your list, where the engine marks the moves that cost you something and the AI coach can explain them. That is what turns a lost game into the next thing to work on.
Questions
Is it free to play against the bots?
Which bot should I play against?
Can I play against bots offline?
Are there bots that play like humans?
Keep training
Take the coach with you
Start free on iPhone, iPad or Android. No ads, and no subscription needed to play, analyse or take the first course.
Sign in and your games, ratings and progress follow you to the other device.
The home screen, on an iPhone. Same app on Android and iPad.