It does quite well on my limited/not-so-scientific private tests (note the tests don't include coding tests): https://aibenchy.com/compare/google-gemma-4-31b-it

It does quite well on my limited/not-so-scientific private tests (note the tests don't include coding tests): https://aibenchy.com/compare/google-gemma-4-31b-it-medium/google-gemini-3-pro-preview-medium/z-ai-glm-5-turbo-medium/ https://aibenchy.com/compare/google-gemma-4-31b-it-medium/go...