Simple Bench
A single-turn reasoning environment based on the SimpleBench dataset, where models are evaluated on their ability to answer multiple-choice questions.
- Domain
- rl-env
- License
- mit
- Published
- Aug 2025
Cite
Notes
Only stored in your browser.
Top score 20.0% by Qwen3 32B - 2 models reporting
Top models
2
Related tools
1Implementations, trainers, datasets and scaffolds linked to this eval.
FAQ
- What is Simple Bench?
- A single-turn reasoning environment based on the SimpleBench dataset, where models are evaluated on their ability to answer multiple-choice questions.
- What is the current top score on Simple Bench?
- The top reported score is 20.0% by Qwen3 32B, across 2 models reporting.
- How can a model improve its Simple Bench score?
- Tools linked to Simple Bench on Sophon include Simple Bench RL Env (Community) - RL environments, datasets, and scaffolds that target this eval.
- What license is Simple Bench under?
- Simple Bench is available under mit.