0

Hangman Agent

Frontier

A dense-reward multi-turn Hangman environment for Prime/Verifiers.

Domain
rl-env
License
unknown
Published
Mar 2026

Cite

Notes

Only stored in your browser.

Attribution

Leaderboard scores
prime-hub
Attribution policy →

Top score 1.55 by GPT-4.1 Mini - 3 models reporting (3 frontier)

Score history

3
00.511.52Jul 24Sep 24Nov 24Jan 25Mar 25GPT-4o-miniGPT-4.1 Mini

Top models

3
Hangman AgentBar chart with 3 bars. Highest value: GPT-4.1 Mini at 1.6.
3 models

Related tools

1
View all

Implementations, trainers, datasets and scaffolds linked to this eval.

FAQ

What is Hangman Agent?
A dense-reward multi-turn Hangman environment for Prime/Verifiers.
What is the current top score on Hangman Agent?
The top reported score is 1.55 by GPT-4.1 Mini, across 3 models reporting (3 from frontier labs).
How can a model improve its Hangman Agent score?
Tools linked to Hangman Agent on Sophon include Hangman Agent RL Env (Community) - RL environments, datasets, and scaffolds that target this eval.
What license is Hangman Agent under?
Hangman Agent is available under unknown.