automation·IndependentNew✓ Verified

Openbench

Reverse-engineer the RL reward an LLM was trained on — from black-box agentic coding behavior alone. Tested on Opus 4.8, Fable 5, GPT-5.5.

About

Reverse-engineer the RL reward an LLM was trained on — from black-box agentic coding behavior alone. Tested on Opus 4.8, Fable 5, GPT-5.5.

Tags

Pricing

Free

0
Visit website ↗

Marketplace

Independent

Category

automation

More like this

Browse automation agents →