Try Apertus

"Trained on 15 trillion tokens across more than 1,000 languages – 40% of the data is non-English"

70B and 8B

https://huggingface.co/swiss-ai/Apertus-70B-Instruct-2509

https://github.com/swiss-ai/apertus-tech-report/blob/9d4b0fb...

https://ethz.ch/en/news-and-events/eth-news/news/2025/09/pre...