Since January 2026
A test suite for over 40 language models
Using a large test suite, we investigate the behaviour of over 40 language models and the quality of their responses depending on:
- different chunking methods
- different CPUs and GPUs
- different prompts on the same topic