[heckraiser]

https://ollama.ai/library/zephyr

I like to point out higher quantization is certainly better (q8 minimally for consistently better results.)