Record · Pokémon Emerald · PokéAgent Challenge (NeurIPS 2025, speedrun track)

Judge's Choice: fewest actions to the first gym (649), 5th by time

Verified Official results published by the challenge organizers.

Last verified: Oct 18, 2026 · Last checked: Oct 18, 2026

Slowest clock, fewest button presses: a good example of why we never merge metrics.

How was this run measured?

Model: Deepest agent (model not disclosed) (source label: “Deepest agent (model not disclosed)”)
Agent: Deepest
Game: Pokémon Emerald
Harness: PokéAgent Challenge Emerald environment (real-time; game keeps running while the agent thinks)
Inputs: per team (documented in the paper's appendix)
Tools: LLM harness
Assistance: none allowed during evaluation runs
Budget: scored on wall-clock time and action count
Status: published
Evidence: PokéAgent Challenge paper (organizers), Sec. 5.2 & App. C (operator)
PokéAgent Challenge: Speedrunning track (operator)
Unknown: underlying model(s) used by the team (not in our source)
Dates: published Mar 16, 2026 · discovered Oct 18, 2026
Changes: Historical record. Nothing to update unless the source changes.

Submit a correction · Data (JSON) · Record ID pokeagent-deepest

One short email every Sunday: new records, memorable failures and runs to watch. Free. Every claim is linked to its source.