A while ago, a 300B model and a 33B model were released but didn't get much attention,
Yesterday, a 120B model was released and is fighting with DeepSeekV4 Pro!?
DeepSWE also exceeded 40 points...??? Qwen3.6 27B also surpassed it
The upstream merge to llamacpp hasn't happened yet, but I downloaded it right away so I can run it as soon as it does.
If the performance matches the score, it seems like it will be a one-hit local LLM.
