I just found out about this new American (San Francisco) model today and I m currently benchmarking it and, so far, it is outperforming my standard go-to model Qwen3.5:122b and actually landing closer to Haiku in my benchmarking test. It s personality is much better than Qwen and it is more naturally creative when it is designing UIs. I have about 6 more tests to go though.If you re interested you can look at my benchmark dashboard here: https://spottedmarley.com/arena/