Benchmark results in the model card: https://huggingface.co/swiss-ai/Apertus-v1.5-70B

Long story short: ~equivalent to Olmo 3.1