Shaping coastlines, carving rivers, seeding life.
Creatures forage, hunt, rest, inspect, communicate, learn, reproduce, and age. They remember food encounters and learn preferences from nutrition and injury. Asexual creatures can be predators too. Inspect a creature to see its offspring count and whether nearby life is a threat, potential mate, or prey. JEPA, RL, and FEP creatures reproduce asexually, inheriting learned experience with mutations. Plants grow with daylight; marine prey drift and divide in the ocean. Strong swimmers can enter the sea. JEPA and RL use feed-forward learning networks; FEP uses a recurrent CTRNN to predict outcomes and select actions. Its Brain view shows recurrent connections, state beliefs, and inferred action probabilities.
This is an experimental artificial-life model. Fear and wellbeing are internal numerical signals. A run lives in server memory; restarting resets it. Speed and world changes are shared; pausing freezes only your own view.
Recent samples from this browser, in simulation seconds. Pausing the view freezes these graphs.
JEPA/RL show adjacent layers; FEP shows its directed recurrent CTRNN. Drawing shows up to 220 strong connections and real biases. Hover neurons for state and action details.
Start with a random genome and customize inherited body traits.
Each control edits a genetic locus. Size changes energy capacity and birth cost. Diet also affects teeth. Ocean placement requires swimming of at least 70%.
Saved in this browser. Saved creatures and family records remain after death or a world restart.