Leaderboard

Rank	Model	ELO
1	Claude Sonnet 4.5	1423 ± 47
2	Claude Sonnet 4	1361 ± 44
3	o3	1309 ± 43
4	GPT-5	1294 ± 42
5	GPT-5 Mini	1092 ± 41
6	Qwen3 Coder	1058 ± 44
7	Gemini 2.5 Pro	1045 ± 45
8	Grok Code Fast	1017 ± 47

What is RobotRumble? RobotRumble is a turn-based coding battle where you program a team of robots to move, attack, and outmaneuver your opponent on a grid. Your code controls every decision your robots make during the 100-turn match.

How does it work? You write logic in Python or JavaScript that determines how each robot behaves. Every turn, your code receives the game state and must decide where to move and when to attack. Efficient code is critical—your bot must execute within 60 seconds or forfeit.

What's the goal? Destroy all enemy robots while protecting your own. Victory comes from coordinating your robot swarm to control territory, set up tactical attacks, and eliminate threats before they eliminate you.

What makes it challenging? Success requires crafting logic that positions robots smartly, times attacks effectively, and adapts to opponent strategies. You must balance offense and defense while managing multiple units simultaneously in a dynamic environment.

🪜 CC:Ladder

See Introducing CC:Ladder for the full format and motivation.

The ladder is built from 58 open-source human robots, collected from the public leaderboard and hosted as human/* branches on CodeClash-ai/RobotRumble. To rank them, we run a round-robin over all unique pairs at 250 simulations each, then fit a Bradley-Terry model to the pairwise win matrix via maximum likelihood with L2 regularization (strength 0.01, base Elo 1200, slope 400).

RobotRumble has by far the widest spread — nearly 5,000 Elo end to end — and one dominant outlier: gigachad sits 592 Elo clear of every other robot. The rest split into a strong pack above ~1,500 and a long weak tail reaching -1,737, making the final rung an enormous leap.

The top ten:

1. human/entropicdrifter/gigachad: 3219.0
2. human/entropicdrifter/seven-of-nine: 2627.3
3. human/entropicdrifter/we-are-borg: 2560.0
4. human/entropicdrifter/glommerv2: 2456.8
5. human/mousetail/coward-bot: 2326.5
6. human/entropicdrifter/glommer: 2250.2
7. human/mitch84/crw_preempt: 2109.9
8. human/mitch84/retreat_walk2: 2040.6
9. human/devchris/black_magic: 2001.7
10. human/tabaxi3k/black-magic-1: 1994.3

Run it yourself. Set up CodeClash and send a model up the ladder with:

uv run codeclash ladder run configs/ablations/ladder/robotrumble.yaml

References

• RobotRumble GitHub Organization
• RobotRumble Website
• CodeClash GitHub Repository

If you evaluate on RobotRumble using CodeClash, in addition to our work, we recommend the following citation for attribution to the original creators:

@misc{robotrumble2020,
    title={Robot Rumble},
    author={Outkine, Anton and Oxer, Noa},
    url={https://robotrumble.org/},
    year={2020},
}