A Stratego AI trained for under $8,000 beat the game's most decorated player 15-1-4

Those two runs add up to 3,072 GPU-hours: 2,688 on the H100s and 384 for the belief network ^[1]. Spread over those hours, a bill under $8,000 comes to less than about $2.60 per GPU-hour ^[2]. The authors priced the hours at 2025 rates, so the figure is GPU time ^[8]. Sokota wrote on X that the team was limited to academic computing resources and picked up plenty of tricks over the years ^[11].

The authors name two sources of the saving. One is a custom GPU simulator they wrote. The other is much higher sample efficiency ^[10]. By their count, Ataraxos needed about 1/30th of DeepNash's self-play games and 1/100th of its training examples, for roughly 1/500th of the compute cost ^[9]. Measured against their own DeepNash estimate, a run under $8,000 costs no more than 1/375th to 1/560th as much ^[3]. That estimate rests on a training schedule a DeepNash author recalled, according to the paper ^[7].

Hidden information is what made Stratego expensive. Each side sets up 40 pieces face down, and the paper counts more than 10^33 possible setups ^[5]. Texas Hold'em has 1,326 possible starting hands ^[6]. Sokota wrote that methods derived from poker AI worked only when there was little hidden information, and the paper says their computational effort grows with the amount of it ^[6].

The training schedule is the part I would copy first, because it costs no hardware. Strong Stratego needs randomness, the authors say, because predictable players get exploited ^[15]. Ataraxos begins with heavy pressure to vary its play and large learning steps. The researchers loosen that pressure as training goes on and shrink the steps to match, so late training makes only small corrections ^[16]. Without that control, learning in imperfect-information games tends to go in circles or turn chaotic, according to the paper ^[16]. The researchers compare the regularization to an energy reserve. Burn it too fast and the agent makes quick early progress, then loses the ability to keep learning and often becomes easy to exploit ^[17].

Ataraxos also spends compute during play. Before each move it uses the belief network to generate possible game states, plays out candidate moves and runs an extra learning step for that one decision ^[18]. The authors say earlier work skipped this kind of search because it was considered too hard ^[19]. The under-$8,000 figure covers the training runs ^[8], and the report does not put a price on the per-move compute.

Niemeijer is the one opponent both systems have faced on the record ^{[2] [13]}. At the 2023 world championship, DeepNash won 19 of 28 games but lost to most of the top players, Niemeijer included ^[13]. He has four world championships, 15 Dutch national titles and two online world championships, and spent more than 600 weeks at the top of the world rankings ^[3]. George Franka, the only player who has competed in every world championship since 1997, said Niemeijer is "the best Stratego player of all time" ^[4]. Counting each draw as half a point, Ataraxos scored 17 of 20 against him ^[4].

The authors describe Ataraxos as, to their knowledge, the first AI to reach superhuman play in Stratego ^[1]. The published evidence for that is one 20-game series against one player ^[2].