Here you can push LLM models to their limits
Ctrl+G is the platform building the world's first Impossible Benchmark: a benchmark designed to make today's models fail.
You can be a part of it by creating real coding challenges.
Your mission is to create tough but fair, real-world problems for LLMs. Once your challenge is ready, we'll test it on a LLM and see if you can outsmart the machine.
→
We currently accept challenges in:
Obvious but worth saying: you cannot use AI to create your challenges
Loading...