A deep convolutional neural network, which takes the whole field as input. The field is split in three layers: the tail, the head and the food. The model uses the Advantage Actor-Critic (A2C) approach. The neural network branches in the last layer: One branch is the actor, it suggests the next step to take (up, down, left, right). The other branch is the critic, it estimates the quality of the current state, i.e., how many points the snakes will be able to collect in the future. The model was trained on 75000 games.